24th May 2020, 9:07 pm | #101 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
I have removed the link L10.
The looks like the TIP36 survived OK, I have a variable voltage supply from about 2 to 11.6v FM and SSB. |
24th May 2020, 9:22 pm | #102 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
This result was to be expected since J3, as we now know, enabled the bias to both FETs. If it is going into self-oscillation the driver can't be involved because as far as I remember you still have LK11 out?
The bias could just be set too high for those replacement FETs. When we had the bias disabled, the radio wasn't eating current so we can at least say that neither of them have failed short-circuit. As you know, I don't like altering factory set settings for any reason but we might have to try that, so, a compromise. When you've tried those things you have in mind and you get to the point where the MOSFETs are either out or disconnected, put the set into TX mode and measure the DC voltages going to both of the pads where the MOSFET gates would be connected. Measure them both as accurately as you can, to at least two digits after the decimal point. According to the diagram, both gates are directly connected together so the voltages on the two gate pads should be identical, but check both anyway. This is so that we can, if we want to, return the bias adjustment to precisely how the factory originally set it. The Bias presets are fed from '8T' which is '8V regulated, transmit only' so it should not matter which TX mode you make this measurement in, but just in case, when recording the gate pad voltages also note down which TX mode you measured them in. In my experience it's quite difficult to test a bare MOSFET with a meter because you can have one on the bench in front of you and you can turn it randomly on and off just by touching the gate. Whatever state the mains cycle is at when you happen to take your finger away, the FET 'remembers' that and stays off, half on, or on. I'll throw your question open to others - how best to test a MOSFET? |
24th May 2020, 9:36 pm | #103 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
I decided to remove the 2 finals for testing, so I could test them on my Chinese multi tester.
They seem OK. But with all this talk about biasing and the 2 finals sharing the biasing, I thought I would take this opportunity to ask advice on which of the IRF 520 devices I have in my spares would be best suited to the job as a matched pair, or are these figures given by my Chinese multi tester pointless. .....C pF.....Vt v a ..1320.....2.05 b ..1320.....2.95 c ...821......3.76 d ...829......3.74 e ...822......3.75 f ...826......3.76 g ...820......3.75 h ...823......3.75 j ..1360......1.95 k ..1360......1.96 l ...1360......1.95 m ...869......3.42 n ...960......3.85 Thanks Mike |
24th May 2020, 9:37 pm | #104 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,903
|
Re: Superstar 6900N V6 - Fault with SSB mode
Quote:
I think the MOSFETs can therefore be tried out in situ. Open J4 and J11 to kill the RF drive. Open the links to both MOSFETs. Insert an Ammeter (something like an AVO on its 10A range in place of the link to one transistor. Play around with that transistor's bias pot and see what you can get its current to do. If you can't control the current and just get lots, then that Mosfet looks bad. Replace both if you have to replace only one. You still want parts out of the same match even if you have got individual bias pots. You still want similar gain. If it seems OK swap the ammeter to the other MOSFET and play with its bias pot. Playing about is an important investigative tool. Make the current go up quite high for a moment and bring it down again. Feel free to make whooshing noises as you do this. If no-one is listening near you it doesn't matter, if someone is, I find it only reinforces opinions they've already formed - so it still doesn't matter. But what you're looking for is odd behaviour. If it seems to jump abruptly in current (The value of the moving pointer ammeter!) and jumps back then something may be oscillating. If the jumps up and jumps down point on the pot are in different places then that's hysteresis and a very strong indication of something hooting But this could be a simple case of gate oxide breakdown if your antenna arcs somehow. With a basic ohmmeter, these MOSFETs have enough voltage rating to take ohms range tests. Gate to source, gate to drain should look open circuit Drain to source (positive lead to drain) should be open if gate is shorted to source, Might look short or open if gate is floating. The same with the ohmmeter the other way round should look like a diode and give a resistance reading, not just zero or infinity. ACHTUNG! the + and - markings on most moving pointer multimeters on their Ohms range are misleading. The Black terminal/lead marked "-" is actually the one driven positive by the ohmmeter's battery. Boy does this catch people out! The reason for this detail is that the way power MOSFETs are made also makes a diode junction between Drain and source. With the transistor powered the normal way, it just sits there reverse biased doing nothing, but it can catch the unwary. The wary are just those who were once unwary David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
24th May 2020, 9:39 pm | #105 | |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
Quote:
|
|
24th May 2020, 9:44 pm | #106 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
Just seen your last 2 posts, I will go and check the bias voltage.
And I dont have the original pairs they got thrown so I cant say. On my last test was the high current on FM and AM also to be expected. Mike |
24th May 2020, 9:45 pm | #107 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
Sorry David, but I think there is a degree of confusion here. Our original plan was to enable each MOSFET one at a time and if necessary adjust its bias preset, but as you rightly pointed out, the radio we are working on now does not have independently settable bias for the output pair.
It was an earlier version (the V3) which did have a split arrangement. I confess I sort of assumed the V6 had the same arrangement because, well, why would it not? Can you answer crackle's question about the best matched pair from those he has? (post #103). |
24th May 2020, 9:50 pm | #108 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
Mike W has mentioned to me a procedure for adjusting the bias, which I was a little nervous about doing. I wonder Mike W if you can post it here for comment.
Thanks |
24th May 2020, 9:52 pm | #109 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
The difference between those last tests and the symptoms you had originally were that originally, when you were in AM/FM mode the transmitter was being fed with a signal, this time it was not as the driver MOSFET was still unpowered. Mike W has a theory that if the output stage is going to be unstable it is more likely to be so when it is not being fed with a signal, so that may be the explanation.
|
24th May 2020, 9:57 pm | #110 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
I'm OK with playing with the bias as long as we have logged the bias voltage setting so we can go back to it if we want to. Take it away, Mike.
|
24th May 2020, 10:00 pm | #111 | ||
Heptode
Join Date: Dec 2014
Location: Limerick, Ireland.
Posts: 901
|
Re: Superstar 6900N V6 - Fault with SSB mode
Quote:
To an extent Enhancement MOSFETs/VFETs don't need the balancing resistors in series with emitters that bipolar do because if one has more current, it will heat and need a higher gate voltage for the same current. However separate bias would be better and obviously the spread is large! I'd pick a couple with 3.7x, likely doesn't matter the last digit. I wrote to Mike: Quote:
|
||
24th May 2020, 10:20 pm | #112 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
As far as the choice of MOSFET pairs are concerned I think I would go for e and h.
Identical, towards the top end VT v and nearly the same c? |
24th May 2020, 10:47 pm | #113 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
It helps when measuring the bias volts to connect J3, oops.
Anyway the Finals bias voltage is AM 3.728v FM 3.728v SSB 3.733v What about the bias for the driver, the driver is also a replacement mosfet. Driver bias voltages AM 4.861v FM 4.861v SSB 3.911v All of the above voltage readings tend to drop slightly (upto 5mv) over a few seconds. Mike |
24th May 2020, 10:59 pm | #114 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,903
|
Re: Superstar 6900N V6 - Fault with SSB mode
When you're running in FM or AM, the gates of the power transistors are being driven hard on to hard off (negative on the gate) The bias voltage doesn't matter much, the transistors don't spend much time being influenced by it. Gross exaggeration, but you get the point.
Now, in SSB, the default condition with a quiet mike is no derive at all to the power transistors. Make some sound and drive appears pro-rata to your audio waveform, and wih phase info that makes it SSB not DSB. So now the bias really counts, it's the biggest game in town. This may be why it made you think it was a mode dependent failing. To test and try power mosfets, you're probably best off building a test jig. A 13.8v power supply via an ammeter and a car headlamp bulb to feed the drain. Neg supply terminal to source, and then a second variable PSU to feed the gate with respect to source. Stick a 1k resistor right on the gate terminal to kill the propensity to oscillate. Have a voltmeter on the gate (well the non-gate end of that 1k) Move the gate voltage to different values and blip the source supply on (push button in series with the 13.8v supply would be good) Log results, chuck at EXCEL and get it to plot. Decide you haven't enough points and add some more. Repeat for other FETs and you should see which ones have similar gate voltages for the same current. Now the wild bit. You only need linear amplification over half the input cycle. The filtering on the output fills in the remainder! So you don't actually need full class-A bias. Sounds wierd, but you have to see it and work it out for yourself before you'll believe it. So find a pair of FETs with the same gate volts for 6A drain current (Guessed figure!) Might need to make that two car headlight bulbs in parallel Don't have power on long. Maybe clip on some heatsinking. Try to keep device temperature constant. One of those jobs where you make it up as you go along based on what's available. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
24th May 2020, 11:01 pm | #115 | |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
Quote:
Now that you have duly logged those voltages I still think those close matched MOSFETs e and h with an identical high VT of 3.75 will be the better choice for the output pair. Put those in, see how it behaves and then we'll ask Mike W to do a walk through for the bias setup. Last edited by SiriusHardware; 24th May 2020 at 11:06 pm. |
|
24th May 2020, 11:03 pm | #116 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,587
|
Re: Superstar 6900N V6 - Fault with SSB mode
Just caught David's last post. David, am I right in thinking that VT = the threshold voltage at which the devices start to turn on? Ideally we need two with matched VT?
|
24th May 2020, 11:54 pm | #117 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,903
|
Re: Superstar 6900N V6 - Fault with SSB mode
More practical to go for something like the operating bias current.
the problem with Vt in data sheets is everyone has a different interpretation of 'just starts' David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
25th May 2020, 6:53 am | #118 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
Hi David
The list of specs in post #103 were measured with a little Chinese multi tester. Would they be good enough or do I need to make a test jig. Would a 2 x 1 ohm resistors in series, and a 6v .3a bulb in series with a 22R, in parallel with the 2R, do as an indicator and current limiter? thanks Mike Last edited by crackle; 25th May 2020 at 6:58 am. |
25th May 2020, 7:46 am | #119 |
Rest in Peace
Join Date: Mar 2011
Location: Basildon, Essex, UK.
Posts: 4,100
|
Re: Superstar 6900N V6 - Fault with SSB mode
Had a reply back from Simon the Wizard.
"Hi Mike. Sounds true in my opinion. I don’t have any bias information sadly. It’s just not my thing. I pay to get stuff done. I’m a modification or firmware guy that’s my area factories / importers use me for sadly. I know in principle on other radios, not on 6900 sadly. I know the bias voltage needs be below I think 3.4v I read it somewhere on a forum. Not much use sadly but not my area Regards Simon |
25th May 2020, 9:26 am | #120 | |
Heptode
Join Date: Dec 2014
Location: Limerick, Ireland.
Posts: 901
|
Re: Superstar 6900N V6 - Fault with SSB mode
Quote:
But I wouldn't bother. The Chinese meter gives close enough matches and they don't need to be identical. Whatever the Vt really is you need to adjust the bias pots in the rig anyway. Just having a similar pair for the o/p is important and not using outliers (FETs with much lower Vt, or much higher Vt), so picking any similar pair between 3V and 4V is likely fine. You should see what the adjustment range is on the two pots with no power feed to the FETs. In any case the Chinese tester definition of Vt isn't important. We can easily pick a similar pair. They don't have to absolutely identical, but I'd need to know the temperature coefficient vs change in Drain-Source resistance, which I don't. I imagine it's less critical than matching BJT or valves, perhaps the designers had a big batch of similar VT, or decided sorting was cheaper than two separate pots for bias. A bad decision. However, EVEN with perfectly matched FETs, it's obvious that no matter what the definition of Vt is, that swapping FETs without realigning the presets is a disaster, worse than changing AC12x or AC18x complementary Ge o/p pair without readjusting bias. We also see that the bias in the rig can easily be too high for some FETs and the Driver bias could be worse for many randomly chosen FETs. As the other poster said, you don't need real Class A bias of exactly 1/2 current at the 6V on AM, the Q of the output load helps maintain the RF current and reduce clipping on negative cycles. However likely the Q and output filtering is too poor for true class C on FM, so simplest approach is to set to max AM power (about 6V) and adjust the DC current to be roughly twice what the RF power suggests, so if 4.5W AM (max is 5W for a rig that does 20W on FM as the FET supply must be 1/2 the FM supply on AM = 1/4 power), the total DC current ought to be less than 2 x 4.5 /6 on the o/p, that's 1.5A (or 750mA per FET but that's irrelevant). So we should set for a bit less than that. Then try for maybe 600 to 700mA on the driver, ideally the minimum with not much distortion. Clipping will give 56MHz if you are on 28MHz. I'd use my spectrum analyser, but a radio that does 56MHz AM would serve. Possibly even look at 4th harmonic on an Airband radio = 112MHz AM. Start at minimum bias and increase to not more than 750mA on the Driver FET for the 56MHz or 112MHz to be a minimum, or a nice 28 MHz sinewave on a scope that does 100MHz. Even a 20MHz scope will make the RF look nice, low pass filter action! |
|