EMRFD Message Archive 5175

Message Date From Subject
5175 2010-09-17 18:08:38 Chris Trask Re: X-Mode Mixer
>
>
>
>
5176 2010-09-17 20:44:40 kb1gmx Re: X-Mode Mixer
5177 2010-09-18 05:00:09 drmail377 Re: X-Mode Mixer
Hi Chris,

As for logic parts, some options come to mind:

1. For HF work and prototype testing the 74AC logic family is fast enough, cheap and fairly easy to find. For example, the latest Softrock SDR's use 74AC74's for I/Q generation. There are logic technologies that are faster, but they're more expensive and (importantly) harder to find IMO.

2. TI makes a bunch of tiny logic parts in SMT, various technologies too but more limited than other package types.

3. Potato Semiconductor (yes that's their real name, they make Potato "Chips") make a line of GHz 74xx logic parts. They retail many of their parts on ebaY. In-fact their Web-store is just a bunch of ebaY listings! They have offices in Taiwan and San Jose, CA. Have a look at some of their GHz bus switches while you are there - just for fun.

www.potatosemi.com

3. Micrel makes some very fast standard logic parts that are quite attractive for driving mixers at VHF/UHF and above.

http://www.micrel.com

For example, Dave (WB6DHW) uses P/N SY10EP52VZG for I/Q generation in his UHFSDR, which you can see here:

http://www.wb6dhw.com/For_Sale.html#UHFSDR

The problem with Micrel is they're notoriously difficult to get parts from. But, Art (K1YK) does the group buys for the UHFSDR and may be able to help. There's a link to Art
5178 2010-09-18 05:17:54 drmail377 Re: X-Mode Mixer
One more thing on logic Chris. You may want to take a look at level translators for the likes of LVDS. LVDS is balanced differential physical interface.

http://en.wikipedia.org/wiki/LVDS

By using something like LVDS you may be able to eliminate at least one transformer(?)

The Fairchild Semiconductor FIN1002 part pops off the top of my head. I use the FIN1002 to convert the balanced LVDS output of a high performance versi
5179 2010-09-18 06:11:00 wa9cgz Re: X-Mode Mixer
I,m considering using the MC10EL89 (SY10EL89) which is a 50 ohm ECL coax driver with two outputs if I understand the data sheets correctly. My idea is to use Dave's WB6DHW UHF project SI570 and ECL dividers for driving external mixers. are these chips a good choice?

Joe wa9cgz
5180 2010-09-18 06:29:21 Chris Trask Re: X-Mode Mixer
>
>
5181 2010-09-18 06:35:36 Chris Trask Re: X-Mode Mixer
>
>I'm watching the development of this "X-Mode" mixer with great interest.
>Please keep us informed.
>
>BTW, when do you expect your diodes to arrive?
>

I'm thinking that they'll be here by Wednesday. Mouser has the BAS40-05V dual common-cathode pairs, just $10 for 100pcs. That will probably be the best diode for now, at least from a cost and availability standpoint.

I've built a small negative impedance amplifier that outputs a 1V peak-to-peak square wave with 5-8nSec rise and fall times. Takes -20dBm of sinewave to drive it. Don't need the ECL logic for the LO driver now.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5182 2010-09-18 06:44:28 Chris Trask Re: X-Mode Mixer
>
>By using something like LVDS you may be able to eliminate at least one
>transformer(?)
>
>The Fairchild Semiconductor FIN1002 part pops off the top of my head.
>I use the FIN1002 to convert the balanced LVDS output of a high >performance version Si570 to unbalanced output. The FIN1003 goes to >400Mbps.
>
>But I think you want a part like the FIN1002 that goes in the opposite
>direction; unbalanced in, balanced out. I don't have a part number
>off-hand for one of these, but I'm pretty sure they exist.
>

Looked it up on Digi-Key. It's the FIN1001. Puts out 0.5V peak-to-peak to a 100-ohm load. I'll get a couple and see how well they work.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5183 2010-09-18 06:48:19 Chris Trask Re: X-Mode Mixer
>
> I,m considering using the MC10EL89 (SY10EL89) which is a 50 ohm ECL
>coax driver with two outputs if I understand the data sheets correctly.
>My idea is to use Dave's WB6DHW UHF project SI570 and ECL dividers
>for driving external mixers. are these chips a good choice?
>

Now that's a nice LO driver. A bit pricey, and Digi-Key seems to be dropping it. Any alternatives, such as a single-function version?


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5184 2010-09-18 07:32:48 drmail377 Re: X-Mode Mixer
Hi Chris, When you say, "I've built a small negative impedance amplifier that outputs a 1V peak-to-peak square wave with 5-8nSec rise and fall times. Takes -20dBm of sinewave to drive it. Don't need the ECL logic for the LO driver now."

Please share with us the "Negative Impedance Amplifier" you mention. I presume you are using a diode biased in the "tunnel" area? Share the circuit you are using with us please. I'm sorry if I'm a bit dense on this subject, I'm a "mixed-signal" pers
5187 2010-09-18 08:09:16 Chris Trask Re: X-Mode Mixer
>
>Please share with us the "Negative Impedance Amplifier" you mention.
>I presume you are using a diode biased in the "tunnel" area? Share
>the circuit you are using with us please. I'm sorry if I'm a bit
>dense on this subject, I'm a "mixed-signal" person - digital and
>analog. One Brain (mine anyway) can only handle so much...
>

Just uploaded it into the files section. There should be an email announcement shortly.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5188 2010-09-18 08:26:08 Blake Michael Re: X-Mode Mixer
A modern day telephone "E6" repeater :)

Michael Blake
k9jri@mac.com





5189 2010-09-18 08:36:40 Chris Trask Re: X-Mode Mixer
>
>>
>> >
>> >Please share with us the "Negative Impedance Amplifier" you mention.
>> >I presume you are using a diode biased in the "tunnel" area? Share
>> >the circuit you are using with us please. I'm sorry if I'm a bit
>> >dense on this subject, I'm a "mixed-signal" person - digital and
>> >analog. One Brain (mine anyway) can only handle so much...
>> >
>>
>> Just uploaded it into the files section. There should be an email
>>announcement shortly.
>>
>
>A modern day telephone "E6" repeater :)
>

Sort of. Just takes -20dBm of sine wave input to get a 1V peak-to-peak squarewave output with 5-8nS rise and fall times with very little if any skew.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5190 2010-09-18 09:24:01 ehydra Re: X-Mode Mixer
Hi Chris -

Do you may have a LTspice project file? Otherwise I will hack it in.

Interesting.

- Henry

--
ehydra.dyndns.info


Chris Trask schrieb:
>>>> Please share with us the "Negative Impedance Amplifier" you mention.
>>>> I presume you are using a diode biased in the "tunnel" area? Share
>>>> the circuit you are using with us please. I'm sorry if I'm a bit
>>>> dense on this subject, I'm a "mixed-signal" person - digital and
>>>> analog. One Brain (mine anyway) can only handle so much...
>>>>
>>> Just uploaded it into the files section. There should be an email
>>> announcement shortly.
>>>
>> A modern day telephone "E6" repeater :)
>>
>
> Sort of. Just takes -20dBm of sine wave input to get a 1V peak-to-peak squarewave output with 5-8nS rise and fall times with very little if any skew.
5191 2010-09-18 10:42:40 Chris Trask Re: X-Mode Mixer
>
> Do you may have a LTspice project file? Otherwise I will hack it in.
>
> Interesting.
>

Nope, sure don't. I don't use computer modeling much except to
determine if something is a good or bad idea. Lots of bad ideas avoided
over the years.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5192 2010-09-18 11:11:47 KB9BVN Re: X-Mode Mixer
Here's how I do it:

No Smoke - Good idea
Little Smoke - Might still be a good idea, think about it some more.
Big Smoke - Idea may be sound but parts selection needs an overhaul.
Big Smoke, Big Fire - NEXT!


----- Original Message -----
5193 2010-09-18 11:20:10 Chris Trask Re: X-Mode Mixer
>
> Here's how I do it:
>
> No Smoke - Good idea
> Little Smoke - Might still be a good idea, think about it some more.
> Big Smoke - Idea may be sound but parts selection needs an overhaul.
> Big Smoke, Big Fire - NEXT!
>

If the prototype works, it becomes a "Proof of Concept".

If the prototype blows up, it becomes a "Poof of Concept". :{b


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5194 2010-09-18 12:09:27 kb1gmx Re: X-Mode Mixer
5195 2010-09-18 12:38:36 Dave - WB6DHW Re: X-Mode Mixer
I have had no problems getting free samples from Micrel. The SY10EP52VZG
and other ECL parts are available from Digikey. The SY10EP52VZGECL outputs
will directly drive 50 ohm inputs. I use the differential outputs to drive the
DBM mixers. The SY10EP52VZG is rated for 4 GHz!

Dave - WB6DHW
<http://wb6dhw.com>

5196 2010-09-18 13:33:22 Chris Trask Re: X-Mode Mixer
>
> I have had no problems getting free samples from Micrel. The
SY10EP52VZG
> and other ECL parts are available from Digikey. The SY10EP52VZGECL
outputs
> will directly drive 50 ohm inputs. I use the differential outputs to
drive the
> DBM mixers. The SY10EP52VZG is rated for 4 GHz!
>

I'd be interested in trying something like this if I could find a line
receiver/transmitter version so I could input a low-level sinewave, such as
the TI SN65LVDS179 but with a higher output capability.

Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5197 2010-09-18 16:21:28 wa9cgz Re: X-Mode Mixer
MC10EL52's on E-Bay 400100908977 low price, anyone have a source
of the MC10EL89"s?

Joe wa9cgz
5198 2010-09-18 16:37:00 ehydra Re: X-Mode Mixer
Hi Chris -

I made a model of your NIA in LTspice but got a few questions:
1. The 2N5770 is in your circuit a PNP, but googled where it is a NPN. I
use the 2N6771 instead. Correct?
2. Operation point. -20dBm on what? 50 ohms? Hm. I changed input so the
output is 1 volts differential.
3. next question is what is the loading at the output?
4. How about the transformer. I guess 1uH all.
5. Just use any rf transistors?
6. The input impedance seems to be very distored!

Please give more details. Anyway, the output transistion time is
simulated as 5-6ns. Not bad.

I uploaded the file in EMRFD group file under ehydra. You do not need
LTspice to see output.

thanks -
Henry

--
ehydra.dyndns.info


Chris Trask schrieb:
>>>> Please share with us the "Negative Impedance Amplifier" you mention.
>>>> I presume you are using a diode biased in the "tunnel" area? Share
>>>> the circuit you are using with us please. I'm sorry if I'm a bit
>>>> dense on this subject, I'm a "mixed-signal" person - digital and
>>>> analog. One Brain (mine anyway) can only handle so much...
>>>>
>>> Just uploaded it into the files section. There should be an email
>>> announcement shortly.
>>>
>> A modern day telephone "E6" repeater :)
>>
>
> Sort of. Just takes -20dBm of sine wave input to get a 1V peak-to-peak squarewave output with 5-8nS rise and fall times with very little if any skew.
5200 2010-09-18 18:19:36 Dave - WB6DHW Re: X-Mode Mixer
Joe:
Digikey has the SY10EL89ZG, which is Micrels equivilent in the SOIC package.

5201 2010-09-18 19:27:39 ehydra Re: X-Mode Mixer
If you like it really fast, you can go CML. For example GigaComm:
http://www.onsemi.com/PowerSolutions/taxonomy.do?id=273
http://www.onsemi.com/PowerSolutions/productList.do?id=262

BTW: The Si570 is available with CML outputs.

- Henry


--
ehydra.dyndns.info


Dave - WB6DHW schrieb:
> Joe:
> Digikey has the SY10EL89ZG, which is Micrels equivilent in the SOIC package.
>
>
5203 2010-09-19 07:07:19 Chris Trask Re: X-Mode Mixer
>
>I made a model of your NIA in LTspice but got a few questions:
>1. The 2N5770 is in your circuit a PNP, but googled where it is a NPN. I
>use the 2N6771 instead. Correct?
>

Typo. Should have been 2N5771.

>
>2. Operation point. -20dBm on what? 50 ohms? Hm. I changed input so the
>output is 1 volts differential.
>

I was using a generator with 0dBm out and a 20dB attenuator. Didn't bother to determine the input resistance.

>
>3. next question is what is the loading at the output?
>

100 ohms.

>
>4. How about the transformer. I guess 1uH all.
>

My usual 4 turns of #32 trifilar on a 43 material 2402 core.

>
>5. Just use any rf transistors?
>

No, they should be complementary pairs so that you have similar dynamic characteristics.

>
>6. The input impedance seems to be very distored!
>

Not surprising, with the positive feedback. A pair of small resistors, such as 47 ohms in series with the inputs should clean it up some.

>
>Please give more details. Anyway, the output transistion time is
>simulated as 5-6ns. Not bad.
>

I was measuring 5-8nSec with no measurable skew. More than good enough to suit my purposes. The only other detail I can offer is that in use the mixer LO input transformer primaries are connected across the squarer output terminals.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5204 2010-09-19 07:20:43 ehydra Re: X-Mode Mixer
Good morning Chris! We have tea time :)

>> I made a model of your NIA in LTspice but got a few questions:
>> 1. The 2N5770 is in your circuit a PNP, but googled where it is a NPN. I
>> use the 2N6771 instead. Correct?
>>
>
> Typo. Should have been 2N5771.

Again typo: I meant the 2N5771. So it is ok.


>
>> 2. Operation point. -20dBm on what? 50 ohms? Hm. I changed input so the
>> output is 1 volts differential.
>>
>
> I was using a generator with 0dBm out and a 20dB attenuator. Didn't bother to determine the input resistance.

Does the generator have 50 ohms output impedance or is it hard-driving?


>
>> 3. next question is what is the loading at the output?
>>
>
> 100 ohms.

I see. Some like an ECL-driver.


>
>> 4. How about the transformer. I guess 1uH all.
>>
>
> My usual 4 turns of #32 trifilar on a 43 material 2402 core.

Never needed the inductance. OK.


>> 6. The input impedance seems to be very distored!
>>
>
> Not surprising, with the positive feedback. A pair of small resistors, such as 47 ohms in series with the inputs should clean it up some.
>
>> Please give more details. Anyway, the output transistion time is
>> simulated as 5-6ns. Not bad.
>>
>
> I was measuring 5-8nSec with no measurable skew. More than good enough to suit my purposes. The only other detail I can offer is that in use the mixer LO input transformer primaries are connected across the squarer output terminals.
>

What is your intended purpose? Your mentioned 40-70MHz LO?

If the squarer is connected to the LO port we don't have 100 ohms load
for the squarer! What makes me curious about if the mixer LO port wants
to see a resistive load for best performance. hm.

I upload a faster kind soon.

Thank you.

- Henry

--
ehydra.dyndns.info
5205 2010-09-19 07:29:42 Chris Trask Re: X-Mode Mixer
>
>> I was using a generator with 0dBm out and a 20dB attenuator.
>>Didn't bother to determine the input resistance.
>
>Does the generator have 50 ohms output impedance or is it hard-driving?
>

50 ohms

>
>>
>>> 3. next question is what is the loading at the output?
>>>
>>
>> 100 ohms.
>
>I see. Some like an ECL-driver.
>

I've been looking at those as the suggestions are posted. Have yet to see a receiver/transmitter that I really like, though the Fairchild STO-23-5 devices are attractive. A bit pricey, though.

>
>>
>>> 4. How about the transformer. I guess 1uH all.
>>>
>>
>> My usual 4 turns of #32 trifilar on a 43 material 2402 core.
>
>Never needed the inductance. OK.
>
>
>>> 6. The input impedance seems to be very distored!
>>>
>>
>> Not surprising, with the positive feedback. A pair of small
>>resistors, such as 47 ohms in series with the inputs should clean
>>it up some.
>>
>>> Please give more details. Anyway, the output transistion time is
>>> simulated as 5-6ns. Not bad.
>>>
>>
>> I was measuring 5-8nSec with no measurable skew. More than
>>good enough to suit my purposes. The only other detail I can
>>offer is that in use the mixer LO input transformer primaries are
>>connected across the squarer output terminals.
>>
>
>What is your intended purpose? Your mentioned 40-70MHz LO?
>

It's just a generic squarer to use for driving mixers up into UHF.

>
>If the squarer is connected to the LO port we don't have 100 ohms load
>for the squarer! What makes me curious about if the mixer LO port wants
>to see a resistive load for best performance. hm.
>
>I upload a faster kind soon.
>

I'll take a close look at that.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5208 2010-09-19 15:40:50 Chris Trask Re: X-Mode Mixer
>
> And the frequency is? Powered with your mentioned squarer circuit?
>

LO is 7MHz, and RF is 10.1MHz and 9.9MHz.

>
> Where is the problem to sample BAS40-05 ? I googled and it is all over
> alround there...
>

I can buy them from Mouser, but I'd like to get a deal on eBay. The
BAS40-04 is on eBay, 100pcs for $3.99 plus $2.50 S/H. Might be worth
trying.

>
> I'm not sure if all sources mean the same chip specs.
> The quadruple is the BAS40-05V and BAS40XY. See:
> http://www.nxp.com/documents/data_sheet/BAS40_1PSXXSB4X_SER.pdf
>

I found you have to be careful who you buy these multi-vendor diodes
from. Vishay makes the worst for capacitance as far as I've seen.
Philips/NXP is best. Recovery times are all over the place.

>
> And for the HP devices:
> http://www.avagotech.com/pages/en/rf_microwave/diodes/schottky/hsms-2805/
> It is not possible to sample them? Maybe they are just to expensive to
> be used final.
>

I'm going to try getting samples of some HSPS-281x devices tomorrow. If
not, I have a few choices on eBay, but I want to wait and see what
configuration works best before taking the plunge.

Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5209 2010-09-19 16:24:09 Chris Trask Re: X-Mode Mixer
>
> I'm going to try getting samples of some HSPS-281x devices tomorrow.
If
> not, I have a few choices on eBay, but I want to wait and see what
> configuration works best before taking the plunge.
>

Went back over my eBay purchased and saw that I had bought 50 pcs of
HSMS-2804 yesterday. That will take care of things for a while.

Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5210 2010-09-21 11:41:27 Chris Trask Re: X-Mode Mixer
The second batch of Schottky diodes arrived this morning, and here are
the performace comparisons:

Diode BAV70 BAT54C HSMS-2804
LO Level 2.0V 1.0V 1.0V
Isolation RF to IF -40dB -47dB -46dB
IF to RF -38dB -47dB -47dB
LO to IF -47dB -59dB -60dB
LO to RF -57dB -67dB -90dB
OIP3 +22dBm +18dBm >+20dBm
Conversion Gain -5dB -4dB -6.5dB

I'm a bit surprised that the conversion gain is lower with the HSMS-2804
diodes. I'm going to get some HSMS-2814 diodes later as they have a much
lower dynamic resistance and junction capacitance.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5213 2010-09-21 12:41:22 Chris Trask Re: X-Mode Mixer
>
> Chris, why doe You use lower LO power for the Schottky diodes? Is it only
that
> they have lower on-voltages??
>

Pretty much. I increase the power until there's no IMD improvement.
I'm not at all surprised that the Schottky diodes take have the voltage of
the original silicon BAV70 diodes.

>
> What happens if the LO power is varied i.e. does the OIP3 increase at
higher LO power??
>

No, it levels off. If you use too much drive power, the IMD performance
deteriorates. That happened around 7V peak-to-peak.

BTW: The best performance with the silicon diodes was with a squarewave
LO, and with the Schottky diodes a sinewave LO worked best.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5222 2010-09-22 11:26:00 ehydra Re: X-Mode Mixer
Chris Trask schrieb:
>> 2. Operation point. -20dBm on what? 50 ohms? Hm. I changed input so the
>> output is 1 volts differential.
>>
>
> I was using a generator with 0dBm out and a 20dB attenuator. Didn't bother to determine the input resistance.
>
>> 3. next question is what is the loading at the output?
>>
>
> 100 ohms.
>

I cleared up the LTspice simulation file and now added the 100 ohms
differential output load you specified.
Problem now: The output is a more or less clean sinusoid!

Input level is 40mVpp. At 300 ohms load there is the transition region
between rectangle and sinusoid output. Higher load resistance gives
rectangle output.

What is going wrong?

- Henry

--
ehydra.dyndns.info
5224 2010-09-22 12:14:14 Chris Trask Re: X-Mode Mixer
>
> >
> >> 3. next question is what is the loading at the output?
> >>
> >
> > 100 ohms.
> >
>
> I cleared up the LTspice simulation file and now added the 100 ohms
> differential output load you specified.
> Problem now: The output is a more or less clean sinusoid!
>
> Input level is 40mVpp. At 300 ohms load there is the transition region
> between rectangle and sinusoid output. Higher load resistance gives
> rectangle output.
>
> What is going wrong?
>

Increase the feedback capacitors slightly.

Chris
5225 2010-09-22 12:47:38 ehydra Re: X-Mode Mixer
Chris Trask schrieb:
>>>> 3. next question is what is the loading at the output?
>>>>
>>> 100 ohms.
>>>
>> I cleared up the LTspice simulation file and now added the 100 ohms
>> differential output load you specified.
>> Problem now: The output is a more or less clean sinusoid!
>>
>> Input level is 40mVpp. At 300 ohms load there is the transition region
>> between rectangle and sinusoid output. Higher load resistance gives
>> rectangle output.
>>
>> What is going wrong?
>>
>
> Increase the feedback capacitors slightly.
>

Between 12pF and 13pF it becomes instable. Below 12pF there is no
change. So this will not work.

You measured it with the scope and 100 ohms differential? Or was the
load the mixer?

- Henry

--
ehydra.dyndns.info
5226 2010-09-22 13:30:16 Chris Trask Re: X-Mode Mixer
> Chris Trask schrieb:
> >>>> 3. next question is what is the loading at the output?
> >>>>
> >>> 100 ohms.
> >>>
> >> I cleared up the LTspice simulation file and now added the 100 ohms
> >> differential output load you specified.
> >> Problem now: The output is a more or less clean sinusoid!
> >>
> >> Input level is 40mVpp. At 300 ohms load there is the transition region
> >> between rectangle and sinusoid output. Higher load resistance gives
> >> rectangle output.
> >>
> >> What is going wrong?
> >>
> >
> > Increase the feedback capacitors slightly.
> >
>
> Between 12pF and 13pF it becomes instable. Below 12pF there is no
> change. So this will not work.
>
> You measured it with the scope and 100 ohms differential? Or was the
> load the mixer?
>

<> Bad solder joint. With a 100-ohm carbon film resistor
soldered properly I still get a fairly good squarewave, and doubling the
emitter currents restores the rise and fall times. In your model, remember
that these film resistors, both carbon and metal, are slightly inductive,
plus I have full-length leads on the resistor. With the mixer as a load,
there will be ldakage inductance in the transformers which will give the
squarewave a bit of an inductove kick.

Chris
5227 2010-09-22 14:42:48 ehydra Re: X-Mode Mixer
Chris Trask schrieb:
>> Chris Trask schrieb:
>>>>>> 3. next question is what is the loading at the output?
>>>>>>
>>>>> 100 ohms.
>>>>>
>>>> I cleared up the LTspice simulation file and now added the 100 ohms
>>>> differential output load you specified.
>>>> Problem now: The output is a more or less clean sinusoid!
>>>>
>>>> Input level is 40mVpp. At 300 ohms load there is the transition region
>>>> between rectangle and sinusoid output. Higher load resistance gives
>>>> rectangle output.
>>>>
>>>> What is going wrong?
>>>>
>>> Increase the feedback capacitors slightly.
>>>
>> Between 12pF and 13pF it becomes instable. Below 12pF there is no
>> change. So this will not work.
>>
>> You measured it with the scope and 100 ohms differential? Or was the
>> load the mixer?
>>
>
> <> Bad solder joint. With a 100-ohm carbon film resistor
> soldered properly I still get a fairly good squarewave, and doubling the
> emitter currents restores the rise and fall times. In your model, remember
> that these film resistors, both carbon and metal, are slightly inductive,
> plus I have full-length leads on the resistor. With the mixer as a load,
> there will be ldakage inductance in the transformers which will give the
> squarewave a bit of an inductove kick.
>

Maybe I'm doing something wrong but I checked all several times. I now
changed the parameters so we get squarewave output.

Input is changed to have -20dBm at 50 ohms. This is 23mVrms.

Coupling coefficient transformer guessed at 0.95 .

Works best if emitter-R is 220 ohms. 680 does not work! At least not
with a resistive only load.

The circuit is VERY sensitive to correct input amplitude. Bad!
Maybe something like ADCMP565 is a better candidate.

I upload the project files and pictures.


- Henry




--
ehydra.dyndns.info
5230 2010-09-22 23:41:22 Chris Trask Re: X-Mode Mixer
> Chris Trask schrieb:
> >> Chris Trask schrieb:
> >>>>>> 3. next question is what is the loading at the output?
> >>>>>>
> >>>>> 100 ohms.
> >>>>>
> >>>> I cleared up the LTspice simulation file and now added the 100 ohms
> >>>> differential output load you specified.
> >>>> Problem now: The output is a more or less clean sinusoid!
> >>>>
> >>>> Input level is 40mVpp. At 300 ohms load there is the transition
region
> >>>> between rectangle and sinusoid output. Higher load resistance gives
> >>>> rectangle output.
> >>>>
> >>>> What is going wrong?
> >>>>
> >>> Increase the feedback capacitors slightly.
> >>>
> >> Between 12pF and 13pF it becomes instable. Below 12pF there is no
> >> change. So this will not work.
> >>
> >> You measured it with the scope and 100 ohms differential? Or was the
> >> load the mixer?
> >>
> >
> > <> Bad solder joint. With a 100-ohm carbon film resistor
> > soldered properly I still get a fairly good squarewave, and doubling the
> > emitter currents restores the rise and fall times. In your model,
remember
> > that these film resistors, both carbon and metal, are slightly
inductive,
> > plus I have full-length leads on the resistor. With the mixer as a
load,
> > there will be ldakage inductance in the transformers which will give the
> > squarewave a bit of an inductove kick.
> >
>
> Maybe I'm doing something wrong but I checked all several times. I now
> changed the parameters so we get squarewave output.
>
> Input is changed to have -20dBm at 50 ohms. This is 23mVrms.
>
> Coupling coefficient transformer guessed at 0.95 .
>
> Works best if emitter-R is 220 ohms. 680 does not work! At least not
> with a resistive only load.
>
> The circuit is VERY sensitive to correct input amplitude. Bad!
> Maybe something like ADCMP565 is a better candidate.
>

After seeing the X-Mode mixer work best with a sinewave LO, I set this
circuit aside. I still test with both the sinewave and squarewave LO
signals just to be certain that I'm not missing anything.

Waiting for another lot of diodes to arrive. I'm also waiting for a
connectorized mixer to arrive to use for testing my idea for small-signal
IMD cancellation. The math all looks good, but the test bench is where the
real proof lies.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5231 2010-09-23 07:06:03 Tim Re: X-Mode Mixer
I'm a little uncertain as to exactly how you are making the sine wave and square wave... but jitter in the squaring process is quite easy to come by and can be introduced by any number of things outside the specs of the parts involved. e.g. power supply ripple or noise. With the ripple-induced jitter at least there are obvious sidebands on an analyzer - when I was working in low jitter clock generation I was never satisfied until I could see the ripple induced jitter on the analyzer, until then I could've just as well believed that all data was fake. I would even purposefully induce ripple jitter just to satisfy myself that the anlyzer could see it if it were there, and to get a ballpark idea of the fundamental slope of dt/dV where V=power supply rail.

I see you critiquing picosecond range specs on Schottky diodes elsewhere in this thread but unless your square wave has less jitter than that then I'm not sure it's relevant.

One of my favorite Tracy Kidder quotes: 'one day during the debugging, his weariness focused on the logic analyzers and the small catastrophes that come from trying to build a machine that operates in billionths of a second. He went away from the basement of Building 14 that day, and left this note in his cubicle, on top of his computer terminal: "I'm going to a commune in Vermont and will deal with no unit of time shorter than a season." '

BTW It's strange to see picosecond range specs on different brands of diodes critiqued in the same post where E-bay floor sweepings are preferred to stuff from an actual distributi
5232 2010-09-23 07:38:53 Chris Trask Re: X-Mode Mixer
>
>I see you critiquing picosecond range specs on Schottky diodes
>elsewhere in this thread but unless your square wave has less
>jitter than that then I'm not sure it's relevant.
>

I've been a bit focused on that spec for the diodes after realizing that the charge storage time of the bipolar transistors in my feedback mixers was the Achilles Heel in the IMD performance.

>
>One of my favorite Tracy Kidder quotes: 'one day during the
>debugging, his weariness focused on the logic analyzers and
>the small catastrophes that come from trying to build a machine
>that operates in billionths of a second. He went away from the
>basement of Building 14 that day, and left this note in his
>cubicle, on top of his computer terminal: "I'm going to a commune
>in Vermont and will deal with no unit of time shorter than a
>season." '
>

Q. Where does an Eskimo have his office?
A. In an ice cubicle.

>
>BTW It's strange to see picosecond range specs on different
>brands of diodes critiqued in the same post where E-bay floor
>sweepings are preferred to stuff from an actual distribution
>chain. Nothing against that E-bay seller.
>

I'm actually being overly cautious about the diodes I buy on eBay. I've passed up offerings of Vishay and other brands of Schottky diodes as they have the worst specs for time and/or capacitance. By way of spec sheets, NXP/Philips come out ahead for the general purpose diodes I've been using. I'm certain that the formal mixer diodes from Avago will end up being the best choice. even though they are not monolithic pairs.

Chris
5233 2010-09-23 07:46:29 Matt Palmer Re: X-Mode Mixer
Re: general purpose diodes I actually find that semtec regularly beats the
penguin snot out of all the datasheets, if I could only get our supply chain
guys to actually buy them.


Matt
W8ESE


[Non-text portions of this message have been removed]
5238 2010-09-23 20:17:37 kb1gmx Re: X-Mode Mixer
5242 2010-09-24 07:10:50 Chris Trask Re: X-Mode Mixer
>
>>
>> >
>> >I see you critiquing picosecond range specs on Schottky diodes
>> >elsewhere in this thread but unless your square wave has less
>> >jitter than that then I'm not sure it's relevant.
>> >
>>
>> I've been a bit focused on that spec for the diodes after
>>realizing that the charge storage time of the bipolar transistors
>>in my feedback mixers was the Achilles Heel in the IMD performance.
>>
>
>I looked at them and considered the CA3046 and friends to be
>antiquated devices. They are good in the realm they are in but
>they are also 40 year old tech and I double the processing
>techniques that have evolved are applied to them as the result
>would not be the same part. I compared them one in the 80s to
>QT2222 a quad set of 2n2222 equivalent devices and the QTs were
>better. Not to say don't use them but they are slow.
>

Of all the CA3xxx devices I had tried over the years, I found the CA3102 to give the best IMD performance. Those are pretty scarce. I later turned to dual monolithic NEC devices, but that product line was too unstable to be considered for long-term.

>
>Mixers are a odd business.
>

Yes, they certainly are.

Chris
5244 2010-09-24 14:41:06 ehydra Re: X-Mode Mixer
I don't think it makes sense to investigate parts so old. A QT2222 is
not found by google.

Is this a typo: 2n5176 ? I think it should be 2n5179 .


- Henry

--
ehydra.dyndns.info



kb1gmx schrieb:
> would not be the same part. I compared them one in the 80s to
> QT2222 a quad set of 2n2222 equivalent devices and the QTs were
> better. Not to say don't use them but they are slow.
>
> In a time when multi-ghz devices are common and generally cheap
> I'd be for looking at them.
>
> I went though trying the Rohde single balanced mixer a while back using 2n3904 in place of 2n5176 after all I was working at 7mhz!
5245 2010-09-24 15:45:52 ehydra Re: X-Mode Mixer
Maybe the charge storage time prob is a manifestigation of symmetry of
the LO waveform driving the mixer?

It would be interesting if IMD goes better if the LO is heavily
bandpassed and then the sinusoid drives the mixer.


- Henry


--
ehydra.dyndns.info


Chris Trask schrieb:
> Of all the CA3xxx devices I had tried over the years, I found the CA3102 to give the best IMD performance. Those are pretty scarce. I later turned to dual monolithic NEC devices, but that product line was too unstable to be considered for long-term.
5246 2010-09-24 16:54:25 kb1gmx Re: X-Mode Mixer