EMRFD Message Archive 11999
Message Date From Subject 11999 2015-12-14 06:09:14 farhanbox@gmail.c... IRF510 : how high? whats the maximum drain voltage one can use with the IRF510?
i have pushed them to 48v. some blew up on the load test at 20 watts (single ended). i am wondering if i can get 100 watts from a pair at 14 Mhz. at 30v I am clocking 30 watts.
- f
12000 2015-12-14 07:05:22 bobtbobbo Re: IRF510 : how high? You can get over 50 watts at 13.8 volts from a pair of IRFZ24Ns. Amp described in June 2010 QST. Have that amp running in my HB SSB xcvr with no problems. I used a regular, larger, heatsink and small low-noise fan instead of the heatsink/fan arrangement used in the article which was much too loud.
Bob, K1AO12001 2015-12-14 07:26:23 Jim Strohm Re: IRF510 : how high? Jim N6OTQAnd of course, going a little farther back in history, remember that television sweep tubes were once favored in lower-end HF rigs and amplifiers because they were cheap, plentiful, and reliable. Times change.Historically, I recall that more than one person has done some casual comparative testing on the various so-called IRF510s available, and all those good folks found a wide range of operating conditions among the casual sampling of devices they used.But nowadays it's one of those devices that if I were to specify it in a product, I'd require a top-tier vendor with full ISO traceability on each delivered device, AND subject 100% of incoming parts to a significantly rigorous in-circuit test -- and reject entire batches if one unit failed.Because of those two issues, you may consider using a higher voltage device with similar input/output capacitance. Please remember that when you began developing your BITX20, the IRF510 was cheap, plentiful, and almost never counterfeited. And at that time it was an excellent, well-reasoned choice.When you get a fake device, all bets are off. When you get a "sorta like, at DC maybe" device like most NTE "replacements," many or most bets are off.Farhan,The datasheet claims 100V, but that's DC and not an SSB signal.Best to test experimentally and stringently document the device mfg. AND the source from which you purchased it, along with your test data. We already know that "IRF510" is more of a guideline and not a specification; because of its popularity in hobbyist apps it is a frequently counterfeited device.
7312002 2015-12-14 08:21:41 farhanbox@gmail.c... Re: IRF510 : how high? jim
i had read that irft10 dies were used by other manufacturers. this device is getting a littl long in the tooth. i dont know of any other device that is as ubiquitous as this one. thus, i continue to experiment with it.
some of mrf devices are so much more attractive. i would love to build mrf150 linear.
- f
------ Original message------
12003 2015-12-14 08:44:29 Lasse Moell Re: IRF510 : how high? Farhan,You may know these... but to me it looks like a true tube replacement!NXP XR family:/Lasse SM5GLC14 december 2015 17:21:37 +01:00, skrev farhanbox@gmail.com farhanbox@gmail.com [emrfd]:
jim
i had read that irft10 dies were used by other manufacturers. this device is getting a littl long in the tooth. i dont know of any other device that is as ubiquitous as this one. thus, i continue to experiment with it.
some of mrf devices are so much more attractive. i would love to build mrf150 linear.
- f
------ Original message------
12004 2015-12-14 09:17:29 iq_rx Re: IRF510 : how high? Hi Farhan,A major variable is class of service, or more specifically the drive to the gate and load on the drain. A tremendous amount of work has been done over the last few decades on high efficiency modes, beyond the old familiar class A, class AB, and class C from our youth. Class E, F and inverse F, and mode J all shape the waveform at the drain for efficiency and sometimes linearity by intentionally loading the drain with specific complex impedances at the fundamental lower harmonics. Those loading networks tend to be low-pass, so sine wave power is delivered to the load, but the waveform at the drain is anything but a sine wave and may have significant voltage excursions.Device robustness is also related to gate drive. Some high efficiency modes tend to imagine the device is either or off, but if the device switches fast enough, has a big drain inductor to the power supply, and a reflective load at the harmonics, then that's a recipe for drain voltage spikes, particularly if you also allow the load at the fundamental to make wild excursions from 50 ohms. One reason the IRF510 was popular for those great 40m class E experiments at Cal Tech was that they were so cheap and easy to replace.It turns out that designing a single-band transistor amp for something like old class AB with a sine wave at the gate and a bigger sine wave at the drain is good for getting reasonable performance and long life out of a solid state device. Turn on the FET with a little quiescent dc bias, provide a drive network that will impress the input signal across the considerable gate capacitance, perhaps with some added resistance to make for a more benign gate load. Calculate the approximate drain load resistance using the classic V^2/2R = power equation, and design a simple network to transform 50 ohms to the calculated R as a starting point. Then adjust at the bench for output, efficiency, and linearity.There are a number of nice more modern FETs for linear SSB service, and those make good amplifiers using circuits in the application notes.Have fun with the experiments,Rick KK7B12005 2015-12-14 09:51:27 Ashhar Farhan Re: IRF510 : how high? rick,your two part paper on linear amplifier design is my standard text. it delineates the metho so simply. i have been experimenting with push-pull narrowly tuned amplifiers of IRF510 class.My regular shack amp is built using those rules of thumb.
Here is the beast :http://hfsignals.blogspot.in/p/25-watt-linear-for-40-and-20.htmlit is interesting to apply link coupled, narrow tuned, air core network to these devices. it works surprisingly well as long as the device side is above 10 ohms. i have pushed them to 60 watts at 28v (Meanwell SMPS cranked up from 24v). I was just wondering if I could tide this arrangment over 100 watts.- f12007 2015-12-14 13:17:45 Graham / KE9H Re: IRF510 : how high? --- GrahamDear -f:Just be sure to be wearing safety glasses when you go for 100 Watts.
==12008 2015-12-14 14:48:00 Tayloe, Dan (Noki... Re: IRF510 : how high? If you really want 100w using these devices, I suggest trying to use four of them. 2x vs. 2x in a push-pull configuration. I have not had a problem getting 7 or 8 w at 12v, so a 24v supply seems high enough to allow 25w per device. With 2x drain voltage peaks, that gives some margin to the device voltage limits.
A push-pull configuration has the nice property that the input C looks to be 1/2 that of a single device. Using 2x devices on each side brings the input C back up to a 1x of a single device.
- Dan, N7VE
12009 2015-12-14 18:21:51 kb1gmx Re: IRF510 : how high? Farhan,Its a very capable device but I find getting the heat out of it the greatest challenge.I'm using the WA2EBY (HF_USE _FET in files under my call) circuit and I get 55W using 28V.At 34V (limit for that supply) its near 75W. FYI at 28V on 20M the input current is about 3.9A(110W). THe device dissipation limit is 43W so 2 of them at 50% you can get to 80W but notfor long due to heating. That power out is with about 2W of drive. Too much drive and theygo bad fast as well. Idle current per device is 100MA with thermal feedback and electroniccurrent limiting. The power supply goes from constant voltage mode to current limitmode at 4.5A (linear supply using LM723).I also used a design for 6M (OZ1PIF) using 4x4 push pull at 34V for 220W. I use a wellmatched antenna.The limit is that the device is a 100V device and IF the signal plus DC exceeds that its gone.Also at 28V ('EBY design) I've found it quite robust but at 34V the die in the package heatsfaster than the heatsink (big!) can absorb it, the to220 case is has high thermal resistance.Also While testing I found that at 34V moderate swr was all that was tolerable (2.5:1)where at 28V I've never failed a device (open, short, wrong filter). I spent a lot of timeabusing it (actually testing it to extremes) before calling it good and putting the screws in.Also by big heat sink I mean a 4x6x.25 inch thick copper plate mounted to a 8x4x2 inch(fins 1.75" tall) aluminum heatsink. During testing to get the heatsink fins to 100F thedevice cases (using IR thermometer) were at 190F (88)C. The actual die by calculationwas over 200C. That hot stability and life is limited. At the time the devices were beingasked to deliver 92W at 34V on 40M, thats 165W DC input or 3.7 time rated power and atmax rated drain current, they failed after 3 minutes. Suggestion stay under 60W powerdevice DC input peak and insure the efficiency is not less than 45%.The key is if the power-out/power-in is less than about 50% the output circuit (transformer)is wrong and your converting DC to heat rather than RF. I did use the BN-43-3312 baluncore as substitutes did not deliver good efficiency and power. However for the secondaryI found that The suggested #20 teflon wire was a good start but I found the secondary ifwound with 7 strands of #30 silver plated Kynar insulated wire (parallel) gave better powerat 10M by 4W (37W out vs 33).Allison12010 2015-12-14 19:49:51 iam74@rocketmail.... Re: IRF510 : how high? This is probably the most sensible approach, one taken by the French BINGO developers (q.v.) among others.
The IRFxxx series has always been a little suspect for RF. Lately it seems to have gotten worse. Check out the IRFZxxx and BUZxx series for good substitutes, especially the BUZ 24 and BUZ44. Again, running 4 or 6 in parallel push-pull makes a lot of sense.
john
Ad5YE
---In emrfd@yahoogroups.com,wrote : If you really want 100w using these devices, I suggest trying to use four of them. 2x vs. 2x in a push-pull configuration. I have not had a problem getting 7 or 8 w at 12v, so a 24v supply seems high enough to allow 25w per device. With 2x drain voltage peaks, that gives some margin to the device voltage limits.A push-pull configuration has the nice property that the input C looks to be 1/2 that of a single device. Using 2x devices on each side brings the input C back up to a 1x of a single device.- Dan, N7VE12021 2015-12-16 14:43:29 kg4ljb Re: IRF510 : how high? Dan,
Perhaps I'm missing something, but I'd think Cin would DOUBLE - when you put FETs in parallel.
Can you clarify?
Thanks - Steve KG4LJB12022 2015-12-16 15:10:12 Tayloe, Dan (Noki... Re: IRF510 : how high? When they are in push-pull (two devices total), the push pull drive makes it look like the Cin of the two devices are in series, thus the driver sees ½ Cin of a normal device. When each side of the push-pull uses 2x devices in parallel (4x devices total), there is 2x Cin (2x devices in parallel on one side) in series with 2x Cin (2x devices in parallel on the other side) so the driver “sees” 1x Cin.
- Dan, N7VE
12023 2015-12-16 18:02:02 Clutter Re: IRF510 : how high? I just recently changed over from 12V to 24V with my WA2EBY amp. I managed to blow
up a pair of IRF510 finals, but it took a good hour of ragchewing on 40CW having
forgotten to turn on the heat sink fan, to do that. My conclusion is that they are pretty
rugged at 24V and 50W output. I do agree with others who suggested four IRF510 for
100W output level- but there are many better choices in RF power FET's today, pretty
cheap, too. BTW remember, in EBY's original article, he points out that efficiency is
'peaky' over the lower end of HF- he got 75W output from two IRF510 on 30 meters,
his best band (at 28V). That same amp produces about 20 watts less on 40 meters,
its least efficient band.
I switched from a 'medium' size heat sink, looks to be aluminum or 'pot metal', perhaps
removed from an old car radio. I'm now running my EBY amp on an all-copper heat sink
from an old Motorola server. It has the same ~4x6x0.25 inch slab of copper like yours,
Allison- but no copper to aluminum interface, it's all copper with soldered or swaged
thin copper fins, lots of them. I'm cooling it with two 40mm server fans. It's hard to detect
any heating at all. Key down for a while produces slight temperature rise on the tops
of the transistors. I've never noticed any perceptible rise in that heat sink. It has 3-4 heat
pipes embedded in the copper slab, FWIW. Perhaps they help. They connect to an
offset area with additional fins. Amazingly, I have been able to buy these on that auction
site for around 10 dollars. It just takes some patience searching out the bargains.
Allison- did the bundle of 7 wires only help on 10m while not hurting the lower bands?
If so, I may try that change to the transformer myself eventually (so far, only 40m
operation here).
73, David K3KY
(Allison wrote:)
Farhan,
Its a very capable device but I find getting the heat out of it the greatest challenge.
I'm using the WA2EBY (HF_USE _FET in files under my call) circuit and I get 55W using 28V.
At 34V (limit for that supply) its near 75W. FYI at 28V on 20M the input current is about 3.9A
(110W). THe device dissipation limit is 43W so 2 of them at 50% you can get to 80W but not
for long due to heating. That power out is with about 2W of drive. Too much drive and they
go bad fast as well. Idle current per device is 100MA with thermal feedback and electronic
current limiting. The power supply goes from constant voltage mode to current limit
mode at 4.5A (linear supply using LM723).
I also used a design for 6M (OZ1PIF) using 4x4 push pull at 34V for 220W. I use a well
matched antenna.
The limit is that the device is a 100V device and IF the signal plus DC exceeds that its gone.
Also at 28V ('EBY design) I've found it quite robust but at 34V the die in the package heats
faster than the heatsink (big!) can absorb it, the to220 case is has high thermal resistance.
Also While testing I found that at 34V moderate swr was all that was tolerable (2.5:1)
where at 28V I've never failed a device (open, short, wrong filter). I spent a lot of time
abusing it (actually testing it to extremes) before calling it good and putting the screws in.
Also by big heat sink I mean a 4x6x.25 inch thick copper plate mounted to a 8x4x2 inch
(fins 1.75" tall) aluminum heatsink. During testing to get the heatsink fins to 100F the
device cases (using IR thermometer) were at 190F (88)C. The actual die by calculation
was over 200C. That hot stability and life is limited. At the time the devices were being
asked to deliver 92W at 34V on 40M, thats 165W DC input or 3.7 time rated power and at
max rated drain current, they failed after 3 minutes. Suggestion stay under 60W power
device DC input peak and insure the efficiency is not less than 45%.
The key is if the power-out/power-in is less than about 50% the output circuit (transformer)
is wrong and your converting DC to heat rather than RF. I did use the BN-43-3312 balun
core as substitutes did not deliver good efficiency and power. However for the secondary
I found that The suggested #20 teflon wire was a good start but I found the secondary if
wound with 7 strands of #30 silver plated Kynar insulated wire (parallel) gave better power
at 10M by 4W (37W out vs 33).
Allison12024 2015-12-16 19:30:27 Clint Re: IRF510 : how high? You might want to look at the HF Packer amp design. It uses a pair of IRF510s running at 24 vdc and puts out close to 35 watts.A common problem with the HF Packer amp is blowing out the finals due to high SWR.73Clint12025 2015-12-16 22:48:43 Ashhar Farhan Re: IRF510 : how high? I get 35 watts with just one IRF510 on 7 MHz at 28v. The EMRFD's IRF530 gets over 50 watts on a single IRF530 at 28v.I guess, the consensus is to use multiple devices to spread the heat sources and keep the drain voltages down but ..."Here be Dragons"1. Higher drain voltages provide better IMD. In our simple quest for power, this is the first causality. Probably paralleling in push pull will provide better performance from these devices.2. Instability. Given that the input and the output are reactive with these devices, The lore that emerged from south indian operators was that the IRF510 linears were inherently instable with 'dipoles'. In turned out that this was symptomatic, unless the linears were well matched, the instabilities set in. 'It works well on the dummy. but with the antenna it goes haywire' was a common refrain.In all, the ubiquitous IRF510s and IRF530s provide interesting challenge. They cost two for a dollar in Hyderabad but I take each blow out very personally (hi!). The other device quite common with the local crowd is the IRF830 that they do push to 100 watts. They just keep a jar of IRF830s by the linear I guess.Most of the interesting work on these devices is coming in from Indonesia. There's an Indonesian facebook builders called Kommunitas Radio BITX. They use these devices well above 100 watts with most ingenious arrangements. Those who are on facebook can take a peek at their projects on https://www.facebook.com/groups/105478459588498/ . I don't follow all of it as most of the discussions are in Indonesian.- f12026 2015-12-17 01:41:08 bubnikv Re: IRF510 : how high? I am getting 110W on 40m from a pair of IRF520's at 24V when driven close to saturation with 3W input CW signal, but I fear the FETs have too high thermal resistance to survive that mode for long. But 4 IRF510's, two and two in parallel, shall be electrically equivalent and if cooled sufficiently, thay shall be reliable.73, Vojtech OK1IAK12027 2015-12-18 06:02:18 Ancel Re: IRF510 : how high? Can't you run a 6-32 machine screw direct to the heat sink thru the
Transistor TO-220 TAB from the top of the board? The FR4 isn't a good
conductor.12028 2015-12-18 07:27:31 Ronan McAllister Re: IRF510 : how high? Just an observation -- those gray colored soft power device insulators (bergquist and pthers) have poor thermal conductivity (from experience) -- even if heat transfer compound is used.
Recommend either a direct connection to aluminum or copper heat sink (this may be undesirable at RF and may require an insulated heat sink) , or, use mica insulator or better, beryllium oxide insulator...
Just my 0.02
Ronan/KB6NHQ12029 2015-12-18 09:05:22 kg4ljb Re: IRF510 : how high? Dan,
Thank you for your reply. I'm still having difficulty with your assessment.
I would think that in order for the input capacitance to halve, the 2 devices in a push pull amp would - somehow - have to be in series. Again, maybe I'm missing something but it seems clear and obvious to me that if anything, they are in parallel, therefore I'd expect C in to double.
In a classic Push Pull amp circuit, The Source connections are usually tied to ground - clearly they are in parallel.
The Gate connections have a low resistance connection between them, afforded by the input transformers secondary. For all practical purposes, they are in parallel.
The drains have a similar situation with respect to the output transformer.
How can the 2 devices not be seen to be in parallel - with resulting additive capacitance's?
For that matter, how can they possibly be seen to be in series?
Sorry to belabor the point.
Cheers and 73 - Steve KG4LJB.12030 2015-12-18 12:02:12 kb1gmx Re: IRF510 : how high? >>>>Allison- did the bundle of 7 wires only help on 10m while not hurting the lower bands?
If so, I may try that change to the transformer myself eventually (so far, only 40m
operation here).<<<<
No cost below 20M and improved above 17M.You found the biggest flaw in the IRF510, that die on a package that can transferheat well would be interesting.FYI the temperature testing was brick on the key at max power (not at -1db compression)It took a good while with that cooling arrangement. Also no fan.I've built many amps for higher and lower power and cooling is everything.Allison12031 2015-12-18 12:06:22 kb1gmx Re: IRF510 : how high? The Packer amp is a packaged design based on the WA2EBY amp, Look at the schematic.The differnce between the two, none save for the packer add an boost power supply to get24V in a 12V system.The thing that will kill the FETs the fastest is over power on the input. 5W will do it if theSWR is over 3:1. At 1-2W I found it hard to cook mine, and I tried.Allison12032 2015-12-18 12:53:44 kb1gmx Re: IRF510 : how high? "Here be Dragons"Indeed!~1. Higher drain voltages provide better IMD. In our simple quest for power, this is the first causality. Probably paralleling in push pull will provide better performance from these devices.HIgher voltages are good to a limit. The limit is a 100V device with 25V power in a high SWR may see 100V peaks on the drain. The other place damage occurs (all power fets) is there is capacitance from drain to gateand if the coupled power and the drive power are high enough the gate punches through.Its a balancing act between over voltage failure and enough voltage to stay in the linear region. I've foundit less a problem at higher drain currents but then cooling has to be good.Last area is marginal bias. A IRF510 without enough heat sink gets hot fast and will thermally take off.Thermally controlled bias and a heat sink to take the heat away is a must.So to help that I use attenuator at the input. For the EBY amp on all bands 2W hit max power or exceeded it.So a 4W radio (some of my HB radios) needs a pad to reduce that to under 2W. I built a pad into the ampso that the drive gets me to 75% power. The result is it also inures the input impedance is goodand the gates are well loaded which helps stability.2. Instability. Given that the input and the output are reactive with these devices, The lore that emerged from south indian operators was that the IRF510 linears were inherently instable with 'dipoles'. In turned out that this was symptomatic, unless the linears were well matched, the instabilities set in. 'It works well on the dummy. but with the antenna it goes haywire' was a common refrain.I've seen this and every case I investigated the device was not well matched to the load. Especially on the gateside. Often the gate is lightly loaded to keep the drive down and the price is poor stability every time.that and poor output transformer design or construction helps the matter. Its still a triode device sofeedback from internal coupling or layout is bonus problems. That leads to a common fault I see,ungrounded heat sinks. The drain and heatsink if not isolated and the sink grounded its a big antennanear a high impedance circuit (the gate side), often with the device standing up. I try to lap the devicedown flat, I cut off the drain lead at the case, use a insulator to the heat sink and a solder lug to attachto the tab going to the output circuit. Makes for a longer layout but the goal is to keep the output far aspossible from the input. Last item on current peaks ground must be ground. If that sounds sillythere's a point there. Most of the board layouts at high current have the FETs mounted with a longSource lead (bad) and the connection to ground is often long (electrically) too, instability, low gain andlow power are often the result.I will repeat it. Most of the power mosfets have gain that can extend to in excess of 100Mhz. Do notlet the high input and output capacitance obscure that as it has limited impact on power generatingability, gain, or power out. There are challenges matching the low input and output impedancesbut its done (even the RF devices like MRF150 has 400pf input capacitance and works fine at 144mhz).>>>In all, the ubiquitous IRF510s and IRF530s provide interesting challenge. They cost two for a dollar in Hyderabad but I take each blow out very personally (hi!). The other device quite common with the local crowd is the IRF830 that they do push to 100 watts. They just keep a jar of IRF830s by the linear I guess.
Funny I've used the device and save for intentional destructive testing I've not fried any. Then againI treat it as I would any device that has gain well into VHF (they do!).I'm running an amp using eight (8) in a 4x4 push-pull. Layout is everything, input matching is also importantand a transformer suitable for 6M (transmission line type no ferrite) and its stable and does 225W at the28V level with good ole IRF510s. The hunk of aluminum under them is impressive and each one is ona ceramic heat transfer wafer as an insulator. Gain for that amp is 12.5DB at 6M! It also idles at 1A(lots of current) so there is 28W of heat just turning it on.>>Most of the interesting work on these devices is coming in from Indonesia. There's an Indonesian facebook builders called Kommunitas Radio BITX. They use these devices well above 100 watts with most ingenious arrangements. Those who are on facebook can take a peek at their projects onhttps://www.facebook.com/groups/105478459588498/ . I don't follow all of it as most of the discussions are in Indonesian.<<<
If you run them class E (switch mode) the IRF510 pair will do a lot more power. Very suitable for AM or CW.The trick with that mode (class E) is that the device is being drive and loaded to the limits of the Rdson andthere are better FETs for that.For real power I just go to the MRF, MRFE, BLF series parts as they can do it and they are cleaner.If $270US is not too scary for one device the MRFE6VP61K part is a kilowatt plus and good to vhf.My preference are in the realm of BLF278 or BLF177 and those do 300-400W with a good 50V supply(and good for 15A!). FYI: most of the RF devices used even push pull tend to want drain current is the300-1000ma even for the moderate 150W devices like MRF150. I mention this as its common to seeIRF510s at 50ma. If you run them at 150-200ma the gain is better and more stable but at 13V thatneeds more than a puny few inch square heat sink.Things I've noted working with power fets and high power transistor amps in general.Allison12033 2015-12-18 12:58:54 kb1gmx Re: IRF510 : how high? "Can't you run a 6-32 machine screw direct to the heat sink thru the
Transistor TO-220 TAB from the top of the board? The FR4 isn't a good
conductor."Cut a window in the board. Use an insulator and thermal transfer greaseplus a shoulder washer with a 4-40 (2.5mm) screw to hould it firm to aslarge a heatsink as practical.The back of the fet is metal for a reason.G10/FR4 has terrible heat transfer properties.Allison12034 2015-12-18 13:09:57 kb1gmx Re: IRF510 : how high? For RF and good heat transfer there are two things available.Mica insulators, thin and good transfer with heat transfer grease.My favorite is a SIO2 ceramic slab about 1mm thick that bothprovides heat transfer and good insulation again with heattransfer grease.Both are inexpensive and stocked by most suppliers. I paid 73 centseach (for 50 pcs) from mouser just recently (532-4170 Aavid thermalloy).Don't forget the shoulder washer to insulate the screw (4-40 or 2.5mm).Allison12035 2015-12-18 13:23:13 kb1gmx Re: IRF510 : how high? >>>The Gate connections have a low resistance connection between them, afforded by the input transformers secondary. For all practical purposes, they are in parallel.<<<
True save for the signal applied to the gates are 180 degrees out of phase. Parallel at DC but at RFanything but. That effectively puts the generator (effective RF source) in series from gate to gate.If you parallel the devices yes that NODE has twice the capacitance. So going from a 1x1 push-pullto a 2x2 push-pull there is a doubling of each nodes capacitance,they are still in series for RF.The whole input capacitance thing is a red herring. Its big to start with and bigger is no worse.In the end you matching an impedance not a simple capacitance.If we take the IRF510 out and use a really nice RF power FET say a MRF150, thats 400PF input capacitance.My 6M IRF510 experimental amp (successful) uses EIGHT in a 4x4 push-pull arrangement. works finewith the right input transformer (9:1 step down) to match the roughly 10ohm gate to gate impedance.Allison