Author Topic: Unobtanium Transistors.  (Read 7390 times)

0 Members and 1 Guest are viewing this topic.

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Unobtanium Transistors.
« on: August 23, 2016, 11:05:46 pm »
I've found a source for some MESFET transistors I *think* I need to repair a sig gen. The datasheets indicate they're almost if not completely identical to the originals but there's a problem.

The company who make them will only supply to ditributors.

Fortunately there's one in my country who are responsive and reasonably helpful but they tell me that the manufacturer will only supply to registered companies.

Now, it's been a while since I had a comapny that would have been acceptable to them so I'm looking for a way around this limitation, does anyone know of a  comapny who would act as a broker and would be willing to order the parts on my behalf?
M0UAW
 

Offline Hero999

  • Super Contributor
  • ***
  • Posts: 10424
  • Country: gb
Re: Unobtanium Transistors.
« Reply #1 on: August 24, 2016, 11:01:06 pm »
Part numbers?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #2 on: August 26, 2016, 10:04:03 pm »
The one I'm trying to source is AM030WH2 but the original part is HWF1681RA, a source for either would be good.
M0UAW
 

Online Cerebus

  • Super Contributor
  • ***
  • Posts: 2818
  • Country: gb
Re: Unobtanium Transistors.
« Reply #3 on: August 26, 2016, 10:55:00 pm »
I've found a source for some MESFET transistors I *think* I need to repair a sig gen. The datasheets indicate they're almost if not completely identical to the originals but there's a problem.

The company who make them will only supply to ditributors.

Fortunately there's one in my country who are responsive and reasonably helpful but they tell me that the manufacturer will only supply to registered companies.

Now, it's been a while since I had a comapny that would have been acceptable to them so I'm looking for a way around this limitation, does anyone know of a  comapny who would act as a broker and would be willing to order the parts on my behalf?

I don't think there's a need to be coy about the companies involved. When companies pursue stupid unhelpful policies like only selling to registered businesses it does good to name and shame them. If I know that a company is going to make it difficult for the individual or small trader to get spare parts I'm not going to specify them in my products. Similarly, if one knows that a distributor has policies in place that stop people getting parts from them, one is not going to enter a distribution agreement with them or use parts that can only be obtained via them. Fortunately your part number gives the game away.

I have some initial sympathy with the OEM, AMCOM, as selling small orders internationally when you're set up to go through distribution is a pain in the fundament and difficult to manage. But it appears that the local distributor would sell you the parts, but have been told by AMCOM not to sell except to registered companies. That's crazy and a fundamental reason to avoid that manufacturers parts like the plague.

If Amcom trip across this, do yourselves a favour - get CJay's address from him and post him a couple of samples. Ditto for APC Novacom, the UK distributor - you're in danger of being tarred with the same brush as your supplier,  talk to AMCOM and get them to see sense.
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #4 on: August 26, 2016, 11:07:32 pm »
I've found a source for some MESFET transistors I *think* I need to repair a sig gen. The datasheets indicate they're almost if not completely identical to the originals but there's a problem.

The company who make them will only supply to ditributors.

Fortunately there's one in my country who are responsive and reasonably helpful but they tell me that the manufacturer will only supply to registered companies.

Now, it's been a while since I had a comapny that would have been acceptable to them so I'm looking for a way around this limitation, does anyone know of a  comapny who would act as a broker and would be willing to order the parts on my behalf?

I don't think there's a need to be coy about the companies involved. When companies pursue stupid unhelpful policies like only selling to registered businesses it does good to name and shame them. If I know that a company is going to make it difficult for the individual or small trader to get spare parts I'm not going to specify them in my products. Similarly, if one knows that a distributor has policies in place that stop people getting parts from them, one is not going to enter a distribution agreement with them or use parts that can only be obtained via them. Fortunately your part number gives the game away.

I have some initial sympathy with the OEM, AMCOM, as selling small orders internationally when you're set up to go through distribution is a pain in the fundament and difficult to manage. But it appears that the local distributor would sell you the parts, but have been told by AMCOM not to sell except to registered companies. That's crazy and a fundamental reason to avoid that manufacturers parts like the plague.

If Amcom trip across this, do yourselves a favour - get CJay's address from him and post him a couple of samples. Ditto for APC Novacom, the UK distributor - you're in danger of being tarred with the same brush as your supplier,  talk to AMCOM and get them to see sense.

To be fair, Amcom state it's for compliance with export regulations which is something I've come across with a lot of American companies, many just will not supply outside the US because it's such a PITA to deal with the government regulations involved

(it did cross my mind to just ask an American board member to get the parts but I'd not want anyone to get into trouble for me by exporting them illegally) but then this is a part of the email from the UK distie:

" please be aware that Amcom will only provide quotations for companies."

M0UAW
 

Online Cerebus

  • Super Contributor
  • ***
  • Posts: 2818
  • Country: gb
Re: Unobtanium Transistors.
« Reply #5 on: August 26, 2016, 11:40:33 pm »

To be fair, Amcom state it's for compliance with export regulations which is something I've come across with a lot of American companies, many just will not supply outside the US because it's such a PITA to deal with the government regulations involved

It's a poor excuse. This would fall under ITAR 'dual use' regulations. I have in the past, as an individual and as a corporate entity, bought parts that fall under that and compliance has never required more than ticking a box on a web site that I adopt a statement that these goods won't be exported to countries or individuals on a named list of restricted destinations/people.

I love the way that in America it's the very politicians who are anti 'big government' and excessive regulation who insist on these regulations and excessive bureaucracy as soon as it might just possibly slightly weaken their beloved military's ability to crush anyone who doesn't toe the American line 100%. Ironically, in the long run it'll probably do more harm to the US export economy than a few bits of electronics in "enemy" hands.
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #6 on: August 26, 2016, 11:53:32 pm »
It's a poor excuse. This would fall under ITAR 'dual use' regulations. I have in the past, as an individual and as a corporate entity, bought parts that fall under that and compliance has never required more than ticking a box on a web site that I adopt a statement that these goods won't be exported to countries or individuals on a named list of restricted destinations/people.

I love the way that in America it's the very politicians who are anti 'big government' and excessive regulation who insist on these regulations and excessive bureaucracy as soon as it might just possibly slightly weaken their beloved military's ability to crush anyone who doesn't toe the American line 100%. Ironically, in the long run it'll probably do more harm to the US export economy than a few bits of electronics in "enemy" hands.
Exactly, if I *really* wanted these for some underhand purpose related to state security or even industrial espionage I wouldn't be approaching an authorised distributor, I'd own a company in the US or have strong links with sympathetic people in the US who did.

As it happens, I'm a hobbyist (in this field) and I got a sig gen from a seller on eBay that wasn't as simple a fix as I'd hoped and want to try to get some value for my money.
M0UAW
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #7 on: August 27, 2016, 08:37:16 am »
I have used these AM030 parts for design work in the past. But this was maybe 12 years ago. I dealt with Novacom in the UK and they were very good at supplying free samples. I've probably still got some of these parts somewhere but I'm not sure if it is OK to release them. I used devices like this for gov/mil design and research work so I have quite a stash of old freebie samples. But they might be ITAR parts.
Try asking Novacom. They might still deal with parts like this?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #8 on: August 27, 2016, 03:57:07 pm »
I have used these AM030 parts for design work in the past. But this was maybe 12 years ago. I dealt with Novacom in the UK and they were very good at supplying free samples. I've probably still got some of these parts somewhere but I'm not sure if it is OK to release them. I used devices like this for gov/mil design and research work so I have quite a stash of old freebie samples. But they might be ITAR parts.
Try asking Novacom. They might still deal with parts like this?

Unfortunately it was Novacom who told me they'd not supply them to anyone but a company.

No mention of ITAR in the document though.
M0UAW
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #9 on: August 28, 2016, 05:55:50 am »
I had a rummage today and all I could find was a very old sample box of AM030 MH4 B1 parts.

These are at least 12 years old and may well be early sample versions of this part when it was first marketed. The labelling on each part looks wonky and blotchy but these are in a genuine AMCOM parts case and they came from Novacom.
I suspect that these are early 'work in progress' samples because the lot number says WIP on the box.

But I'm not sure any of this matters as these aren't AM030 WH2 parts they are the AM030 MH4B1 variant. My parts here are 28V, 6GHz parts.

See the image below:

I'm not sure about the WH2 variant as this is a very fast part and could prove to be very unstable as a drop in replacement. It still has a lot of gain at 12GHz. Did IFR recommend it? If so then I guess it will be OK.

I'm pretty sure I don't have the WH2 variant here as I was working on PA driver designs up to about 3GHz back in those days.

The AM030 parts I have here look very similar on the outside but the technology inside is different. I think my part is good to 6GHz in a typical design but it's highly unlikely to work as a drop in part for your sig gen.
« Last Edit: August 28, 2016, 07:16:22 am by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #10 on: August 28, 2016, 08:03:02 am »
One thing to note is that the s parameter data for these parts is very different across all three parts.

eg the
AM030 WH2
AM030 MH4B1
Hexawave 1681RA

I downloaded the data for all three parts and tried to do a simple simulation using a basic feedback circuit to flatten the gain from 0 to 3GHz. These parts behave very differently in a circuit like this even though they appear similar on the datasheet. If I set up a circuit using the Hex part that gives a flat (unconditionally stable) gain of about 10dB across 0 to 3GHz then if I swap to the AMCOM parts the circuit shows a lot of gain peaking near 3GHz and the circuit is no longer (unconditionally) stable. I'd expect it to show signs of oscillation.

Do you have the original schematic for the sig gen? Does the original MESFET part use resistive feedback to flatten the gain?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #11 on: August 30, 2016, 06:37:25 pm »
Unfortunately I don't have a schematic so I will need to reverse engineer one.

IFR don't appear to supply anything other than a module level troubleshooting guide for £75.

 I had to ask twice before they acknowledged the existence of the 3413 I have.

I can't seem to find a supplier for the original Hexawave parts, there's not a lot of proof that Hexawave even exist any more so the AM030 parts were suggested by another forum member, to my untrained eye they looked similar enough to try.

I don't know where else to look now if they're not suitable and will probably have to accept I've wasted a large amount of money.

 

M0UAW
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #12 on: August 31, 2016, 06:29:38 am »
OK, don't give up just yet...

IMO there are at least three possible scenarios

1/ IFR will (eventually) help you with a repair but it will probably cost more than you paid for the generator. So I guess this scenario is unrealistic.

2/ Your Hex part(s) are dead and IFR won't help you at schematic level and you need to find/fit an alternative part without them. This isn't game over, but it does mean that the path ahead will be the most interesting in terms of getting an alternative part and doing an effective repair.

3/ There is a chance that your Hex part(s) is/are not dead. These parts need a negative bias or they will bias/go to hard conduction like a crowbar switch and cause overheating in the devices themselves or the PSU that supplies this part. i.e. if you somehow end up with zero volts bias at the gate the part will hit the Idss figure in the datasheet of about 1A and it will not be good for 'something' in your sig gen. Either the Hex part will fail through thermal stress or the PSU will overheat or limit. If it is the latter then the devices themselves may still be OK and you could have an issue with the lack of a negative supply.

A common failure mode for devices like this is gate to source breakdown/leakage inside the device itself which pulls down the negative gate bias by itself. i.e. the breakdown leakage current itself will be enough to pull down (remove) the existing negative bias and the device will then go into self conduction as if the negative bias was not present. This means the part is damaged and needs to be replaced.

Gate to source breakdown could be caused by a sudden burst of RF (or low frequency) instability in the device itself during normal operation. This can cause a large RF waveform at the gate and this causes internal breakdown due to overvoltage.

You mentioned that you want two of these MESFETS. My initial fumblings with a simulator showed that the Hex part performs best in terms of gain flatness (LF to 3GHz) and stability (up to many GHz) if the part is run in a 25 ohm system rather than 50R.

Does your sig gen use two of these parts in parallel or push pull?

The simulation also suggests that there should be some resistive feedback from gate to drain in the form of a series resistor and (DC blocking) capacitor.
Is there such a network fitted across the top of the device itself? The simulation predicts that this needs to be a very short/direct path hence the need to piggy these parts rather than fit them to one side on the PCB.

With this all in place and some series resistance in the gate of a few ohms the simulator predicts that the device delivers a flat gain of about 10-11dB from LF through to 3GHz.





« Last Edit: August 31, 2016, 06:47:02 am by G0HZU »
 

Offline KE5FX

  • Frequent Contributor
  • **
  • Posts: 835
  • Country: us
    • KE5FX.COM
Re: Unobtanium Transistors.
« Reply #13 on: August 31, 2016, 06:57:05 am »
3/ There is a chance that your Hex part(s) is/are not dead. These parts need a negative bias or they will bias/go to hard conduction like a crowbar switch and cause overheating in the devices themselves or the PSU that supplies this part. i.e. if you somehow end up with zero volts bias at the gate the part will hit the Idss figure in the datasheet of about 1A and it will not be good for 'something' in your sig gen. Either the Hex part will fail through thermal stress or the PSU will overheat or limit. If it is the latter then the devices themselves may still be OK and you could have an issue with the lack of a negative supply.

Another 'gotcha' with these depletion-mode parts is that they will make it look like the drain power supply bus is shorted, from the perspective of a simple ohmmeter check.  I remember making that mistake when I first encountered them in an HP 8663A.

Quote
With this all in place and some series resistance in the gate of a few ohms the simulator predicts that the device delivers a flat gain of about 10-11dB from LF through to 3GHz.

How much output power is required?  If it's < +20 dBm and < 3 GHz, there are plenty of other ways to skin that particular cat.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #14 on: August 31, 2016, 07:11:41 am »
Quote
How much output power is required?  If it's < +20 dBm and < 3 GHz, there are plenty of other ways to skin that particular cat.
My concern here is that the PCB circuit and layout will be optimised for the Hex part and its package.

Also, the sig gen will typically need a fairly meaty part here in order to meet its linearity specs when used in IQ mode. So I would expect to see something rated at a watt or so but run in backed off mode. Also, the circuit may have some frequency compensating/stabilising RC parts dotted around the Hex part.

Also, I suspect that the circuit will include an active (negative) bias network for the Hex part that typically settles at the required drain current when the gate voltage is steered somewhere in the ballpark of -1V.

I don't know how much of a working window this active bias network will have but it may only work in closed loop across -0.6V to -2V for example. So this narrows down the choice of equivalent parts unless this circuit was modded to widen the working window of the active bias.

The AMCOM 030 MH4 devices I have here look to be a 'drop in' part in terms of the active bias requirements. If I'm allowed to subtly change the resistive D-G feedback resistance and the series gate resistance I can make the AM030 MH4 device give the same flat 10-11dB RF gain response to 3GHz whilst maintaining unconditional stability (according to the simulator and the supplied s2p models that is)
So it wouldn't be a drop in part but it may just require a few changes to resistor values in the existing IFR circuit. However, the AM030 MH4 is a 28V part so the s parameter model won't be accurate when used at the lower supply voltage of the Hex part. So there is some risk here in terms of getting the circuit to give the right gain response when run at a lower supply voltage.

Of course, the ideal scenario would be to try and get some samples of the original Hex part.
« Last Edit: August 31, 2016, 07:21:45 am by G0HZU »
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #15 on: August 31, 2016, 03:56:20 pm »
Hopefully there's a picture linked in to this post of the important bits. (the board was in rather a mess when I got it, there's been some damage to other parts of it which I've repaired but not shown and the MESFETs were covered in black flux residue, there was a copper shield over both parts which I have lifted to identify them)

https://s19.postimg.org/6q2yvuzwj/20160713_201836.jpg

The 2 LT1086 regulators got *very* hot after some short period of time, when I started looking at this sig gen I disconnected the supply from the H1681 devices at L49, L48 and the regulators stayed cool.

A simple multimeter test does indeed indicate a DC short so I really need to check that negative supply is good and working, I think that should be on TP82 and TP83.

I've not drawn out the circuit yet but I think the devices are being used push-pull, T8 at the top right of the image drives the MESFETs, T9 next to L48.

If it yields a working signal generator I am happy to modify component values but wholesale circuit changes I'd prefer to avoid.

My unit does not have an attenuator so the output level should be 0dBm to +22dBm, currently it's considerably lower and rather distorted.
M0UAW
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #16 on: September 01, 2016, 06:01:31 am »
OK, the image makes sense to me in that the circuit uses a couple of baluns to form a push pull amplifier and the devices look to be at 25ohms each. There's some series gate resistance formed from the two parallel resistors R321 R322. I'd expect these to make up an equivalent resistance of a few ohms. I used 6 ohms in my simulation here.

There's also what looks to be an RC pre emphasis on the output. Again, I'd expect this to be a few ohms in parallel with a few pF. When you removed the top copper foil 'shield' from the devices did it form part of an RC network? i.e. was it mounted on a small resistor and a capacitor in series?

I'd expect to see RC feedback fitted across the top of the device and not just some form of copper shield.

If you look at the top of your image you can see a similar push pull circuit. This uses a couple of old Stanford (RFMD) SHF-0189 devices (marked as H1 on the top of the device) and you can see that there is a series RC 'feedback' circuit across the input to output. I have a lot of design experience using this Stanford part! The RC feedback network is used to give a flat gain response over a wide frequency range. It needs to be done with care or the device can become unstable especially if the resistor in the feedback has high self capacitance.

My advice would be to check out the negative bias circuit for the Hex parts. There may well be a very similar active bias network used with the little SHF-0189 devices. These too will go into hard conduction and will overheat if their individual (active?) negative bias circuit fails.
« Last Edit: September 01, 2016, 06:05:22 am by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #17 on: September 01, 2016, 06:10:47 am »
If you want to see a typical active bias circuit and how it is designed see this old Stanford/Sirenza/RFMD application note.

http://application-notes.digchip.com/147/147-47819.pdf


I have used this type of bias circuit with the SHF-0289 in the past and it means you don't have to manually set the bias for the device during production testing. The drain current is defined by the circuit components in the active bias circuit.

There is a slight penalty wrt large signal linearity with an active bias network but I doubt the Hex parts will be run very hard in your case. Have a look to see if there is a circuit like this associated with the SHF-0189 parts and also your Hex parts.

If you find that the Hex parts use an active bias system I can show you how to troubleshoot it safely to prove it works OK.
« Last Edit: September 01, 2016, 06:17:34 am by G0HZU »
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #18 on: September 01, 2016, 08:02:27 am »

There's also what looks to be an RC pre emphasis on the output. Again, I'd expect this to be a few ohms in parallel with a few pF. When you removed the top copper foil 'shield' from the devices did it form part of an RC network? i.e. was it mounted on a small resistor and a capacitor in series?
It's not visible in my picture but there's  a chip capacitor on the gate to the shield on both devices, on the drain there's a chip resistor to the shield, on TR2 there's a cpacitor and resistor stacked in parallel to the shield.

If you look at the top of your image you can see a similar push pull circuit. This uses a couple of old Stanford (RFMD) SHF-0189 devices (marked as H1 on the top of the device) and you can see that there is a series RC 'feedback' circuit across the input to output. I have a lot of design experience using this Stanford part! The RC feedback network is used to give a flat gain response over a wide frequency range. It needs to be done with care or the device can become unstable especially if the resistor in the feedback has high self capacitance.

My advice would be to check out the negative bias circuit for the Hex parts. There may well be a very similar active bias network used with the little SHF-0189 devices. These too will go into hard conduction and will overheat if their individual (active?) negative bias circuit fails.

The negative bias is present on the SHF-0189s (all of them) they're hovering around -4V on the gates.

The bias for the HRF devices is missing, it's hovering around +0.27V for both. I think it's coming from a pair of identical LM358 based sub-circuits on the underside of the board but it's too late this evening to trace it out.

It's not going to be particularly easy to work out what's going on with those LM358s as I'd need some extension boards or cables to allow me to check voltages, I'm thinking the chips are cheap enough to test by replacement but first I need to be certain they're the source of the gate bias.

All my power supply voltages check ouy on the test points on the IQ board so I'm confident the fault is on the RF deck.
M0UAW
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #19 on: September 01, 2016, 08:21:51 am »
Quote
It's not visible in my picture but there's  a chip capacitor on the gate to the shield on both devices, on the drain there's a chip resistor to the shield, on TR2 there's a cpacitor and resistor stacked in parallel to the shield.
Ok, that's good.

Can you post up a decent closeup image of just the HEX stages together?

I need to see if there are any printed shapes that add capacitance. The simulator predicts that there is an advantage to fattening up the microstrip (beyond 25R) at the device input as this helps boost the response towards 3GHz and make it easier to get a flat gain response.

The RC parts associated with the shield are also there to flatten the gain. At low frequencies this delivers negative feedback and this tames/flattens the gain down at low frequencies and helps with the input impedance and stability here. Without this D-G feedback the device gain skyrockets at low frequencies. The two little resistors R321 R322 in series with the gate help with input matching and stability too.

Up at higher frequencies the shield RC (drain-gate feedback) parts will tend towards positive feedback due to parasitic phase shifts in this RC network and also within the Hex part itself. It looks like someone tried very hard to minimise the path inductance here and they used a fat piece of copper to give a very direct RC connection between gate and drain. But there will still be some phase shift towards positive feedback at the top end and this helps boost the gain response up at 3GHz. But this has to be done with care to avoid instability.

Also, the feedback resistance used here needs to have very low self capacitance in its packaging  or the circuit can go unstable up at several GHz. So they probably used carefully selected resistor packages here.

If the Hex parts are dead I think it will be possible to make the AM030 MH4 part work here but it will mean a change to a few of these critical resistors. Otherwise it will have too much gain peaking and may be unstable. The AM030 WH2 variant would be much more difficult to tame in this circuit as it has a lot of excess gain across the whole range and up to many GHz.


« Last Edit: September 01, 2016, 08:31:36 am by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #20 on: September 01, 2016, 09:41:22 am »
If the missing negative bias is due to the opamp circuit rather than the hex part pulling down its own negative bias then I guess it might be worth wondering why they used opamps here. Could the negative bias circuit be quite complex and versatile?

Opamps are great for summing things so maybe this sig gen can dynamically alter the bias depending on the sig gen mode. Eg cw signals vs IQ signals.
So there could be a DAC involved or even some missing lookup table data. But this is all just speculation.

I think you ideally want a backed off class A amplifier here for best linearity but maybe it can alter the bias towards class AB for cw or FM signals up towards +20dBm?


« Last Edit: September 01, 2016, 09:50:41 am by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #21 on: September 02, 2016, 11:07:40 pm »
Quote
The negative bias is present on the SHF-0189s (all of them) they're hovering around -4V on the gates.
Sorry, I misread this the first time. Are you sure you are measuring this at the gate of the device? I assumed at first you had measured the negative supply from which the bias is derived.

A gate bias of -4V on the SHF0189 devices would put the part into pinchoff. Normally, you would expect to see about -1.2V at the gate if the device was biased for approx 100mA drain current. If there really is -4V at the gate then the devices are pinched off and this may be deliberate in that the stage may be deliberately disabled by the sig gen or maybe there is a bias control fault on this amplifier stage too. A -4V bias seems really odd unless IFR sometimes disable this stage with this method. Maybe the stage isn't needed for some modes or maybe they deliberately bias it 'off' in order to improve the 'RF Off' performance when you turn off the RF on the front panel.

Most sig gens break the signal path and set the output attenuator to >>100dB attenuation to disable the RF output because this is quick and easy but maybe your sig gen also mutes the SHF0189 parts by setting them to -4V pinchoff.

Does the SHF-0189 gate bias go to -1.2V if you turn on the RF output from the sig gen?
« Last Edit: September 02, 2016, 11:19:26 pm by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2449
  • Country: gb
Re: Unobtanium Transistors.
« Reply #22 on: September 02, 2016, 11:41:32 pm »
Also, I would expect to see some opamps used as part of the ALC system and these would be located fairly near the output amplifier stage. Could these opamps be the ones you have spotted on the PCB underside? Maybe these are ALC opamps and nothing to do with the amplifier bias?

I'm not sure I can offer much more advice/help than this because (without a service manual or schematic or access to the hardware) I'm currently guessing everything.
« Last Edit: September 02, 2016, 11:43:11 pm by G0HZU »
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2472
  • Country: gb
  • M0UAW
Re: Unobtanium Transistors.
« Reply #23 on: September 06, 2016, 12:27:03 am »
Ah, seems my last post before my weekend away didn't post. I'll re-send it later with pictures attached.

I think you're right about the -4V on the SHF0189s if that's cutting them off hard, there seems to be more than one signal path, i suspect it's switching between different paths depending which 'range' is required. I'll run it through a few tests to confirm that.

The op-amps could well be ALC as the gate is connected to a diode on the underside, RF sampling perhaps.

Once I've charged the battery for my 'big' digital camera I'll get some better pictures and post those to show the op-amp circuit, the HEX stages and will also test the negative bias on the preceding stages, with and without RF on and in different frequencies to see if they change.

Thank you for all the help, BTW, I'd be rather lost without it.

M0UAW
 

Offline rickells

  • Contributor
  • Posts: 29
  • Country: ca
Re: Unobtanium Transistors.
« Reply #24 on: January 02, 2017, 01:41:15 pm »
hi,

I have an Aeroflex IFR 3416 6 GHz Signal Generator that appears to have the same problem as CJay's 3413.

I have No RF output, an ERROR # 522 (Power supply problem on the RF board) and high heating.
I have just started to try to understand the issues this week, but am quite lost without schematics.
The manufacture will Not supply the Service manual nor schematics.
My unit does appear to come with the options 003, 005, 006, & 021 (electronic attenuator, dual Ch. ARB source, fast puts mod & CDMA waveforms).

I have been trying to follow CJay's & others postings, but, (please forgive my humorous plight) rather like Tolkien's Elven folk, the posts keep popping up with a different thread heading (was difficult to find if not Aeroflex 3413 )!

Can someone please supply a rough sketch of the Aeroflex 3413 output stage so that I can attempt to follow it ?
I am going to try to find some H1681's to repair my unit, but am wondering it if is now also the bias supply (OpAmps ?) to the O/P RF finals ? 
Somewhat confused, clarification via a schematic would help.

Can CJay & others please keep me posted as to repairs & parts suppliers ?
Is there anyplace one can find the H1681 or HexWave 1681RA ?

Hoping to repair my Aeroflex IFR 3416 for my ham amateur radio EME project.
thank you !
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf