EMRFD Message Archive 11212

Message Date From Subject
11212 2015-06-20 18:45:47 lmeeny Tayloe Mixer Difficulties

I've been experimenting with the Tayloe mixer for nearly a year but am almost ready to throw the towel in. I've solved the initial problems, mostly my errors and misunderstanding how the circuit worked. For the last several months I've been unsuccessful understanding and correcting the last problem.

The circuit's plagued with low frequency, 10 to 20 KHz, rail-to-rail oscillations at the outputs of the I/Q summing amplifiers that follow the 4 channel buss switch.  The circuit will run properly for several 10's of seconds to several minutes.before the oscillations begin building until they reach the upper and lower output voltage limits of the amplifiers, LTC1677's.

I would be happy to share my schematics and PCB layout drawings with anyone with applicable experience.



11213 2015-06-20 19:15:50 Russell Shaw Re: Tayloe Mixer Difficulties
11214 2015-06-20 20:57:01 lmeeny Re: Tayloe Mixer Difficulties

Thanks for the quick reply.

The amplifiers just now aren't connected to anything, not leaving the PCB. All they drive is a X10 probe and feedback resistors.

11215 2015-06-20 21:29:12 Dana Myers Re: Tayloe Mixer Difficulties
11216 2015-06-20 21:42:33 Russell Shaw Re: Tayloe Mixer Difficulties
11217 2015-06-21 04:39:18 Steve Dick Re: Tayloe Mixer Difficulties
Hi Ed. The oscillation problem may be layout related or load related.  Circuitry around the summing node should have leads as short as possible.  The output should not be driving a highly capacitive load.  If the layout is etched in stone and leads are too long around the summing node, one approach is to add a small capacitor from summing node (negative node) to output. This adds a pole in the closed loop response (low pass filter) but is actually adding a zero in the open loop response, which may mitigate the oscillation.  You could start with, say 100pF.  You don’t want it too large as you really don’t want to effect the closed loop bandwidth much.  If a highly capacitive load is causing the problem,  another way to mitigate this problem is to add a small resistor in series with the the output node and move your feedback resistor to the output node, so the resistor decouples the effect of load capacitance on the open loop response.  This will also result in a low pass filter at the output with an RC time constant of the output resistor in parallel with any load resistor  and the load capacitance.
“Digital Steve”, K1RF with an analog response!!
11224 2015-06-21 13:29:50 lmeeny Re: Tayloe Mixer Difficulties
Steve and All,

Thank you for the quick responses. I've uploaded schematics and PCB layout to the files area. I apologize for not creating a folder to hold all the files. First time upload and I just messed up.

You all have given me lots to chew on. I'll report back as I progress.

Thanks VERY much again.

11227 2015-06-22 03:12:01 g8asg Re: Tayloe Mixer Difficulties

I think you've written the solution on the circuit diagram:
You've assumed the mixer provides a 50 ohm source impedance for the amplifier. In fact the amp sees a 0.22u capacitor so its gain will rise with frequency to the point where it oscillates. You must limit the gain for all frequencies with resistors between the mixer capacitors and op-amp inputs. I use 100 ohms and 10K feedback.

11228 2015-06-22 04:36:20 k1rf_digital_stev... Re: Tayloe Mixer Difficulties
Hi Ed. I looked at the schematic and layout. First, he 74CBTLV3253 analog switch is only rated for 2.3 V to 3.6 V but it appears you are operating it at 5V which will likely damage the switch.  You need to fix this issue before you fix the oscillating op amp issue. . The 3.3V going to the input transformer should really be set for the middle of the range of the analog switch's corrected power supply voltage.
As far as the op amps, I don’t like a capacitor hanging directly on the summing (-) node of the op amp. I have seen this in 1 or 2 of the original Tayloe articles and it is theoretically fine but not necessarily from an op amp stability standpoint.  A cap on the summing node is adding a pole to the open loop response – not a good thing.  You can visualize this by mentally breaking the input to the summing node. The transfer function consists of the op amp's transfer function and the transfer function of the feedback components driving the summing node.  You have an R-C driving the summing node which is introducing a pole in the open loop response at a relatively low frequency.  A small cap from summing node to output should help this.You then have a pole-zero transfer function which helps.  Even better is to insert a small resistor, say 10 ohms, in series with the summing node keeping the feedback resistor right at the summing node.  This will further help with stability along with the cap across the feedback resistor, say 100-200 pF as a starting point.
Let me know how you make out.

"Digital Steve", K1RF
11235 2015-06-22 22:19:31 lmeeny Re: Tayloe Mixer Difficulties
Thanks for the observation. I hadn't thought of it that way but it makes sense now that you mention it.

Another problem was pointed out to me by Rick. I was using a 3.3V version of the quad buss switch but operating at 5.0V, not good. It's being replaced with a TTL input version FST3253.