Author Topic: How to deal with thermal instability of transistor bias current?  (Read 4291 times)

0 Members and 1 Guest are viewing this topic.

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
I built this MOSFET-based push-pull RF amplifier, but used IRF530s because IRF510s were not readily available.

Then I proceeded to adjust the bias current of the amplifier and things started going wrong.

I was shooting for a bias of about 50 mA through each FET. However, the power dissipation from this bias is enough to warm up the FET a bit, making the bias current highly unstable -- with a peak around 50 mA -- above this, the current tends to increase slowly (over minutes) as a kind of thermal runaway, and below this, it seems to decrease.

Is there a good way to solve this problem? Or am I approaching it wrong?
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #1 on: April 14, 2018, 06:36:39 am »
I’ve built a couple of RF PA’s with the IRF510. Depends entirely on your topology. Push pull in that is going to get hot. I can only get 10W out in class C with tiny heat sink and barely getting warm. Big heat sink IMHO and see where you get. I can bias mine up to 500mA without terrible problems but I’m running 13.8v only.

Also gate capacitance of IRF530 is much higher so matching network may need to be redesigned.

I’d try and find some IRF510’s actually. You can get them from Tayda I think.

« Last Edit: April 14, 2018, 06:42:34 am by bd139 »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: How to deal with thermal instability of transistor bias current?
« Reply #2 on: April 14, 2018, 06:52:30 am »
Use a Vbe multiplier, or a thermistor in the voltage divider, to set bias.  Thermally couple the active device to the main heatsink.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: How to deal with thermal instability of transistor bias current?
« Reply #3 on: April 19, 2018, 06:27:26 am »
OK, so I have this thing amplifying, but I've managed to burn up at least two IRF530s. Is the problem that I'm not using IRF510s, or something with my circuit?
 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #4 on: April 19, 2018, 09:57:55 am »
The problem is not necessarily the substitution but the basic design.

As Tim said (indirectly) you need to have something in the circuit that compensates for the change in device characteristics with temperature - this design has none. There's a hint that the original designer perhaps should have their soldering iron licence revoked:

I destroyed many IRF510 FETs during testing. In fact I blew a small hole in one and another into several pieces. It was quite a shock when the first one was destroyed because it made a loud noise like a rifle being fired.

Once I got tired of replacing the FETs, I built a current sense circuit, which shuts off the bias once the amplifier draws more than about 3 amps from the PSU. I think this circuit is essential.

Dealing with bias instability by turning the power off is (*cough*) not a recommended technique.

This PA really needs a redesign - I'm not an RF guy so please don't expect one out of me. Perhaps someone else can help you salvage this, but I've a sneaking suspicion that this is one where starting again with a reliable, well designed, well tested circuit design might be the wiser option.
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3233
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #5 on: April 19, 2018, 10:04:08 am »
OK, so I have this thing amplifying, but I've managed to burn up at least two IRF530s. Is the problem that I'm not using IRF510s, or something with my circuit?

Have to used a Vbe multiplier or thermistor based bias circuit as suggested?  Without some form of bias compensation the circuit is fundamentally flawed, it will never be stable.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #6 on: April 19, 2018, 10:25:22 am »
You don't need a Vbe multiplier or thermistor bias on RF push pull amplifiers. They aren't the same as audio amps.

It's really quite common to blow up switching MOSFETs when you're throwing RF amps together. Problem here is that real RF MOSFETs are quite bias stable, but they cost 40x the amount of an IRF510. If you're wiling to lose a few then it's a good trade off.

Honestly I've not had any problems with bias stability myself. The issue that caused bits of MOSFET to fly at me has always been drain impedance mismatch, overdriving the gates, VHF oscillation (which you can't see easily but can pick up with a field strength meter) which blows the buggers up. Even construction and layout is potentially an issue.

Big thing here is the MOSFETs are going to be in linear region a lot of the time as this is not a class C or E amp so bolting the thing to a massive chunk of metal is the first step. The original author's design is pretty standard.

I'm almost 100% certain this is VHF osc rather than a bias issue.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: How to deal with thermal instability of transistor bias current?
« Reply #7 on: April 19, 2018, 12:48:03 pm »
You don't need a Vbe multiplier or thermistor bias on RF push pull amplifiers. They aren't the same as audio amps.

Well...

Hmm, it's an interesting intersection of generalizations.  Also, I'm unsure if you are applying the case of the following paragraph or not:

Quote
It's really quite common to blow up switching MOSFETs when you're throwing RF amps together. Problem here is that real RF MOSFETs are quite bias stable, but they cost 40x the amount of an IRF510. If you're wiling to lose a few then it's a good trade off.

In essence you've constructed a polychotomy of eight possibilities:
- Is RF-transistor based or not
- Needs thermal compensation or not
- Is AF amplifier or not

To which of the eight were you referring? :)  Because what works for some of those cases, will not work for the others.

And, some methods can be applied to all eight cases, with unequivocal success.  One should probably strive for the latter, rather than the former, no?

Quote
Honestly I've not had any problems with bias stability myself. The issue that caused bits of MOSFET to fly at me has always been drain impedance mismatch, overdriving the gates, VHF oscillation (which you can't see easily but can pick up with a field strength meter) which blows the buggers up. Even construction and layout is potentially an issue.

Potentially?  Construction and layout is always paramount, especially to that last one (oscillation)!

Quote
Big thing here is the MOSFETs are going to be in linear region a lot of the time as this is not a class C or E amp so bolting the thing to a massive chunk of metal is the first step. The original author's design is pretty standard.

"Standard" doesn't mean good.  Public schooling is awful in many ways, yet standard.  Television programs are insipid to vapid, but are standard.  Bad drivers are standard.  A bad "Joule thief" circuit is standard, for whatever reason.  And it should come as no surprise that bad RF amplifier circuits are also standard.

Standard just means enough people have repeated it to some suitably arbitrary degree of completion and function, that they shared it, without having realized its real problems, and perhaps deciding not to share it after all.  The Darwinian spread of an idea -- a meme.  Remember, a Darwin award is only won if reproduction has not happened.

It doesn't have to be good, indeed it often seems like there is mimetic fitness to sub-par designs.

Perhaps it is because the designs are simple enough to attempt, yet frustrating enough to be memorable?

The sheer popularity of something should, at best, be seen as cause to inspect it critically.  Because the chances are, no one else has done so!

Anyway, back to the circuit: there isn't even a shunt resistor for setting the two bias currents!  What the hell -- how are you even supposed to -- what?!  They could've at least added source resistors, with the bonus of source degeneration also reducing distortion.  Regarding AF amps, there are many "shallow class AB" and class A amps that run very hot, and are either designed as self-biased single-ended followers (a few Pass amps are this, IIRC), or always have thermal compensation (pretty much every conventional complementary follower).  The only difference between such circuits, and RF amps, is the use of transformers for AC coupling and impedance matching is rather more feasible at RF, and one must pay attention to high frequency transistor properties (gain rolloff, capacitance, inductance), which has the main effect of excluding large NFB loops from the design.

HTH,

Tim
« Last Edit: April 19, 2018, 12:50:44 pm by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #8 on: April 19, 2018, 01:22:12 pm »
That was an awfully written post I agree! Sorry!

In summary one should strive for being cost effective in such circumstances as a mistake is better to cost the average ham a couple of IRF510's then some more expensive devices.  Regarding "Standard doesn't mean good", yes I agree but the priority here is:

1. works, because what is the point otherwise.
2. doesn't spew harmonics out.
3. cheap
4. good

In that order. Sometimes the design never gets to (4) unless you work for Yaesu or Icom etc.

I am considering building this as I have the parts floating around and doing some analysis on the design. Maybe a month or so off though due to workload.
 

Offline CopperCone

  • Super Contributor
  • ***
  • Posts: 1415
  • Country: us
  • *knock knock*
Re: How to deal with thermal instability of transistor bias current?
« Reply #9 on: April 19, 2018, 01:47:40 pm »
I never worked on an rf amp other then a nice integrated mmic.

How come composite amplifiers are not used, like a feedback with a fast cfb opamp? So the gain is controlled in relation to the bias voltage change?

Or are there parameters effected by bias that cant be controlled with simple feedback and tuning like distortion change over bias voltage? Or does it completely destroy span?
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: How to deal with thermal instability of transistor bias current?
« Reply #10 on: April 19, 2018, 05:10:43 pm »
May or may not apply, but I've built class A audio amps with the IRF530 and found that with a relatively huge heat sink, they are bias stable. Above a certain level, forget what it is, the tempco is negative and all is well. Below that level it's positive. If the heatsink is large enough, you can stay in the stable region at a reasonable temperature.
 

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: How to deal with thermal instability of transistor bias current?
« Reply #11 on: April 20, 2018, 06:31:37 am »
Well, yesterday I played around with this thing a ton, and actually learned about it:

0. This is definitely an oft-repeated amp. I think a big part of it is that it's 1. very simple *and* cheap and 2. can realistically be tuned without having access to an oscilloscope. I do note that this design is a big different from the "original" -- no inductors, and uses a tapped transformer rather than a series balun for the input. It does make some harmonics; for ham radio usage it requires a low pass filter.

1. I suspect that not-all-that-much-educated amateurs have a strong and not totally rational aversion to amplifiers that don't seem push-pull or otherwise symmetrical. I know that many people have successfully used this amplifier to communicate.

2. I don't think I actually destroyed any mosfets completely, and only partially destroyed one of them. One issue is that each mosfet needs to cool completely after overheating before it will behave properly.

3. I need (and bought) better heat sinks. And a fan.

4. I'm a bit worried my JDS6600 will not make quite enough drive for low frequencies, may need to make the input transformer bigger. Or get a pre-amp.

5. Something is wrong with my input transformer, only one side of it is outputting anything. (the other outputs a very weak and noisy sinewave). Need to break it down and remake, ideally with wire of more appropriate thickness and better trifiliar technique.

6. I actually intend to change this thing to FQP13N10's after I get my Mouser order made in order to take advantage of suggestions in this website: http://m0rzf.co.uk/20W_Amplifier/index.html. I see the linked website has diodes as temp compensators.

7. As shunt resistor, you mean a current sensor resistor? I just used an ammeter.

8. Any advise on VHF oscillation? Should I be able to pick it up with my DS1104Z?

My goal essentially is to generate sine waves at around 20-50 watts and 1.5 to 3 MHz.

 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #12 on: April 20, 2018, 07:50:01 am »
VHF oscillation should turn up on that scope. If you get trouble, stuff a 10-22 ohm resistor in series with the gate and add more decoupling capacitors.  That seems to kill off any problems I had with similar amplifiers.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: How to deal with thermal instability of transistor bias current?
« Reply #13 on: April 20, 2018, 08:13:07 am »
6. I actually intend to change this thing to FQP13N10's after I get my Mouser order made in order to take advantage of suggestions in this website: http://m0rzf.co.uk/20W_Amplifier/index.html. I see the linked website has diodes as temp compensators.

Some quick excerpts to hopefully trigger your skepticism:

Quote
TO-220 packages have inductive wire bonds

Maybe they do, maybe they don't.  The leads are ten times longer and therefore higher in equivalent inductance -- you might want to worry about those first!

Quote
Switching FETs hotspot badly and the bias point is unstable… OnSemi agree, Infineon agree, Microsemi agree…

A false generalization -- the bias point is only unstable in the region where it is unstable.  This region is smaller for some types than others.  It is always present.  Some types just burn up (wholesale) before reaching it.

All you must do is simply choose a device which has an SOA which includes all available bias conditions*.

*For any angle of the trimpot, too, mind.  It would be a crummy circuit indeed that cannot save itself from the errant turn of the technician's screwdriver, especially an untrained amateur!  I hate suicidal circuits, and call them out every time I see them!


Quote
I found these are most suitable in terms of medium gate capacitance, low-ish transconductance and good thermal conductivity:

The choices listed are all newer types -- calling them "low-ish" is peculiar, because newer types have more Gm/Id than ever before.  This can make them even more prone to runaway!  We must depend on other design characteristics of these parts, to obtain good linear operation.

Curiously, the single most powerful development, SuperJunction technology (applicable for high voltage MOSFETs, over ~300V I think), is apparently free from breakdown.  If not by nature, then by additional design steps, that seemingly all manufacturers have followed.  It sure is nice, however it's happened -- high voltage transistors are higher performance, and more affordable, than ever!

The article is hardly bald-faced-wrong, no -- but it is also nowhere near something to believe unconditionally.  Likewise the articles referred to, from the manufacturer's mouth even, are far from infallible (the On Semi article in particular lacks critical context in an early paragraph: when talking about PTC, they're referring specifically to the resistive saturation (triode) region only, not the linear (FET "current saturation") region; this is clearly distinguished in the Infineon article, however).


Quote
8. Any advise on VHF oscillation? Should I be able to pick it up with my DS1104Z?

Yes*.

*For the IRF510, most likely.  For the STW13N60M2, maybe not.  Those types tend to sing at 200-400MHz, with the cause usually being a G-S capacitance that instead forms a tuned-gate resonant circuit with the lead inductance.  Again, TO-220 works against you, severely.

The usual fix for such oscillations is a ferrite bead on the gate lead, but that's only practical at low frequencies, or repeat rates (SMPSs are, in a sense, pulsed RF amps, with a typical repeat rate in the 100s of kHz to low MHz).

Well, if we're only talking a few MHz here anyway, that's not a problem.  Maybe an 11m amp would cook the ferrite bead (from gate drive power) though.  (Note: ferrite beads don't "cook off", they just thermoregulate around Tc, as that causes their added impedance to drop to nil.  So, such an amp might be okay at key-down, but a couple seconds later, you get UHF whistlers!)

It's quite practical to build an amp with SMTs -- a two layer PCB with 2oz copper and lots of vias can draw 5 or 10W out of a PDSO-8 sized part, and some quite beefy transistors come in such packages.  (Although the opposite case is usually the problem: that you're looking for smaller transistors, so that capacitance is small; but not so small that the power dissipation is pitiful.)  Packages like these (DFNs and such) have essentially no internal lead length, beyond what you can see, giving you maximal control over parasitics.

Not that DFNs are the easiest thing for an amateur to build, but leaded SMTs (like DPAKs and SOICs) are available in all sorts, at modest expense to parasitics or power dissipation.  People need to realize that SMTs are actually easier; heck, it almost seems like coddling to keep putting THTs out there where they're not necessary!


Quote
My goal essentially is to generate sine waves at around 20-50 watts and 1.5 to 3 MHz.

Hardly a challenge -- these can be synthesized digitally at high efficiency, with a switching converter.  Assuming you didn't need modulation on that, that is (but then, that wouldn't be a sine wave?).  :-//

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #14 on: April 20, 2018, 08:29:21 am »
All you must do is simply choose a device which has an SOA which includes all available bias conditions*.

*For any angle of the trimpot, too, mind.  It would be a crummy circuit indeed that cannot save itself from the errant turn of the technician's screwdriver, especially an untrained amateur!  I hate suicidal circuits, and call them out every time I see them!

You could silkscreen "Turn fully clockwise to catch fire" next to the trimmer ...
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14071
  • Country: de
Re: How to deal with thermal instability of transistor bias current?
« Reply #15 on: April 20, 2018, 08:51:57 am »
Thermal compensation is a real option, though the amplifier might still be stable with low bias in a kind of deep class B region - but with higher distortion. Thermal compensation is not really complicated. Just replace the 78L05 with something like a VBE multiplier or a string of diodes, that are in reasonable thermal contact to the heat sink.

The IRF510 is an old type and still has a reasonable useful SOA. With more modern SMT types one has to keep an eye on the SOA.
 

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: How to deal with thermal instability of transistor bias current?
« Reply #16 on: April 21, 2018, 06:50:46 am »
Quote
Quote
My goal essentially is to generate sine waves at around 20-50 watts and 1.5 to 3 MHz.

Hardly a challenge -- these can be synthesized digitally at high efficiency, with a switching converter.  Assuming you didn't need modulation on that, that is (but then, that wouldn't be a sine wave?).  :-//

Tim

I indeed am not modulating them -- they will be variable power, but that's going to be quasistatic.

I was not aware that there were such things as DDS synthesizers that directly output over 20 watts of power. I'm using a JDS6600 to provide my sinewaves, and then trying to amplify them with this thing. What would you do?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: How to deal with thermal instability of transistor bias current?
« Reply #17 on: April 21, 2018, 07:17:27 am »
A half-bridge is essentially a 1-bit power DAC.  Alright, calling it "synthesized" is a bit of a stretch from the traditional use of the word (RF or musical synthesizers).  Put together gate drivers, power MOSFETs and bypass capacitors, and avoid capacitive load conditions.  Easy enough to get kW, and the efficiency kills anything else.  If you need low distortion, you will need a fairly high order filter (note that the inverter has a very low output impedance, so use a one-port-shorted prototype for the design), but that shouldn't be a big deal.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: How to deal with thermal instability of transistor bias current?
« Reply #18 on: April 30, 2018, 12:38:29 am »
Well, I charged ahead. Got some actual IRF510s and fixed some kind of (possibly capacitive) short to ground.

Now I am able to get some actual output. Unfortunately, at higher power outputs things look the attachment. Yellow is the output (50 ohm dummy load at the end of clip leads) and blue is one of the FET gates relative to ground.

My guess is that I don't have the bias set very well. Or maybe I just need a higher input voltage.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: How to deal with thermal instability of transistor bias current?
« Reply #19 on: April 30, 2018, 06:41:40 am »
It’s going to have some distortion. Need an LPF on it.

 

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3166
  • Country: au
Re: How to deal with thermal instability of transistor bias current?
« Reply #20 on: May 01, 2018, 10:36:48 am »
A bias control cct I developed for an audio amplifier some years ago. Same basic principles should apply for a linear RF amp.

"I am going to try a variation of the bias voltage circuit I suggested in another thread a while back that has 2 pots, 1 for offset and the other for slope. Put the slope at zero, and set the offset for the right quiescent current with the amp cold. As it heats up the current will increase, so adjust the slope pot that set the right amount of downward slope of mV per deg C till the current gets back down to where it should be. Finally when the amp is fully hot, tweak the slope pot till the current is exactly right. Now the bias should more or less track the temperature over the whole range. The only error will be if the diode forward voltage tempco or the fet threshold voltage tempco is not a straight line. Actually the threshold voltage is tempco slightly curved so with a bit of luck the diode curve will be too. We'll see. Anyway, it should be a whole lot better than that ugly hit-and-miss "amplified diode" setup using a single transistor and a pot. Anyone still using that setup in this day and age should be put in jail."
 

Offline ikraseTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: How to deal with thermal instability of transistor bias current?
« Reply #21 on: May 01, 2018, 07:17:35 pm »
That looks more complex than the whole amp.

I actually did not have any bias runaway problem with the irf510s.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf