| razirafi |
| Hello: I just finished repairing my USA204 amp by replacing the MoSFETs and the PWM (SG3524) chip in the SMPS section. The original design uses the BUZ11 but since I have about eight RFP50N06 lying around in my parts bin, I used them instead since they have a much higher rating. They differ in their internal drain to source diode. The RFP50N06 uses a zener diode. The problem now is that the turn-on delay is about three minutes:eek: . After checking the schematic (courtesy of Sounstream Technologies), I probed pin16. when power is applied this pin starts out at about 10V and gradually decreases to about 1.3V, then the amp turns on. I replaced C8 (100uF/16V) with a 1uF/16 and R9 (22.1K) with 150K. Now the delay is about eight seconds. I tested the value of the original resistor and it is 22K exactly. I tested the amp in bridged mode for about half a day and no problems with the sound although the chassis was quite hot. Now when I apply power the power-on LED slowly lights up to full brightness instead of an instant-on condition after the delay. Can anyone give an explanation for this turn-on delay behavior? |
|
|
| razirafi |
| Sorry its not pin 16 but rather pin 10, the shutdown pin. And no the amp does not get quite hot, it gets real hot with the MoSFETs giving off the most heat. |
|
|
| Perry Babin |
I think the 3524 may have seen a production change. I've seen a couple of amps that required a 10k ohm resistor between pin 8 and 10 to return the delay to something near the intended delay. If I'm not mistaken, I didn't change the delay cap.
The power supply FETs should not get as hot as the output transistors (generally BJTs in the soundstream amps I've seen). If the power supply transistors are getting hotter than the output transistors, there's likely a problem with the drive circuit for the power supply transistors. |
|
|
| Bertje |
I think you better put BUZ11 in instead of the 50N06's. I have bad exprience exchanging these MOSFETS without changing the rest of the circuitry. (other amps ..)
Rob. |
|
|
| razirafi |
| Ok thanks for the insights. Well since R8 was also fried I replaced it with an off-the-shelf 10% tolerance type. Could this be a source of drive imbalance to the swithchers? |
|
|
| Perry Babin |
The 10% tolerance resistor may be exactly 75 ohms. You should check it to confirm that it's not too far off. If you have several, check them all and use the one that's the closest to 75 ohms.
As was suggested, it's best to use the original part number as replacements. You always want to reduce the number of variables when troubleshooting a circuit.
Can you measure the operating frequency of the power supply? Sometimes you can have a shift in value of the timing components (resistor or capacitor) and the operating frequency will be too high or too low. |
|
|
| djQUAN |
| quote: | Originally posted by razirafi
Sorry its not pin 16 but rather pin 10, the shutdown pin. And no the amp does not get quite hot, it gets real hot with the MoSFETs giving off the most heat. |
maybe there's too little deadtime because of the higher gate charge.......is there an unusually high input current even at idle? |
|
|
| razirafi |
| I will check the idle current, operating frequency, and take some photos of the waveforms. |
|
|
| razirafi |
| I just opened the amp again today to take some measurements and I found out that the input filter capacitors were blown (C2,C3). I replaced them and measured the idle current (no input signal and no load speakers connected) to start out at about 7Amps, gradually settling down to about 6.3Amps. The input filter capacitors then were getting warmer and warmer. Attached are photos of the waveforms. |
|
|
| razirafi |
| waveform (expanded) |
|
|
| Perry Babin |
What were the vertical amplifier and timebase settings on the scope?
Were the probes on the gate or the drain? |
|
|
| razirafi |
| Vertical setting was on 5V/div (probes on X10) and time base was on 10uS and 5uS, respectively. Probes were connected to the PWM pins 11 and 14 (emitters 1 and 2) before the gate drive resistors. |
|
|
| Perry Babin |
If the scope is accurate, that's ~60kHz. That's almost double what you'd find in other amps. It's likely that the timing cap or resistors is out of tolerance.
Read the values on the timing cap and resistor and use the formula on the datasheet to see what it's supposed to be running at with those values. |
|
|
| razirafi |
| You are correct. My approximate calculation is about 56kHz. If the frequency is higher does it follow that the output voltage will also increase or will the duty cycle compensate for the frequency increase to keep the output voltage at the set value? Could this also be the reason for the rise in temperature and subsequent blowing of the caps? |
|
|
| Perry Babin |
An increase in frequency won't necessarily change the output voltage.
What could be happening is the FETs are not being turned off quickly enough. If this is the case, you'd have both banks of FETs on for a short period of each cycle. This could lead to excessive current draw and excess ripple current for the input filter caps. This could be the reason they're heating up.
If you monitor the voltage on the gate side of the gate resistor, you'll likely see that one bank begins to turn on before the other turns off (voltage falls below ~3.3 volts). If the gate waveforms from opposite banks are both above ~3.3 volts at any point in time, it's a problem. Generally, you'd want to see the gate voltage drop to below 1 volt on one bank before the other bank begins to turn on. |
|
|
| razirafi |
I checked the waveforms directly at the gate terminal and they were the same as when checked on pins 11 and 14. By the way the actual waveforms were in-phase, I just inverted the other channel to view them symmetrically:confused: . Aren't they supposed to be out-of-phase? The datasheet of the SG3524 gave the formula for the frequency as f=1.18/RtCt. With the values in the schematic (Rt=1.9k and Ct=.0047uF), f=132kHz. In push-pull configuration it is effectively about 66kHz.
I will try to remove the gate drive from one bank of FETs and see what happens. |
|
|
| Perry Babin |
Don't try to run the amp with only one bank connected. The other bank will likely fail.
The outputs have to be out of phase. Double-check your setup. Place both probes on the same point. They should give precisely the same trace.
The waveform should have been slightly different on the other side of the gate resistor but with 10 ohm gate resistors, it may be difficult to see the difference.
I have an SA245 here and it has a similar supply. The attached image shows the waveforms on the gates. The timebase is at 5uS. The vertical amp is at 5v/div.
Can you post a photo of the waveforms on the primary windings?
This amp has a 1.91k ohm timing resistor and a 4.7nF timing cap. For some reason, it's oscillating at 53.35kHz instead of the calculated frequency of 66kHz.
I also noticed that the amplitude of the traces is different for your amp. Is that a calibration error or is it the true amplitude? |
|
|
| razirafi |
| I see. I just ran the amp without the drive to one bank of the FETs for about two minutes:eek: . The amp did turn on and the filter caps did not get hot to the touch although they warmed up a bit. Placing both probes on the same point gave precisely the same trace. The different amplitudes of my traces could be accounted for the vertical position setting or a calibration error. I'll run the test again with another scope to check the amplitudes and also post some photos of the primary winding waveforms. We do have the same values for Rt and Ct. Thanks for your help so far. I've learned a lot from this discussion:) . |
|
|
| Bertje |
Just compare the output of the SG35xx with and without the MOSFETs. I think the other type (other than original) of FETs are the cause.
Rob. |
|
|
| Perry Babin |
Bertje:
The difference in the input capacitance of the FETs is generally what causes the problem. The originals have a Ciss of ~1500pf. The replacements have a Ciss of ~2000pf. The transistors (MTP50n05E) in the amp I have here have a Ciss of 3000pf. I paralleled a 1000pf cap across the gate and source and the current draw at idle was still very low (0.863A).
This was a surprise... I changed the value of the timing resistor to increase the power supply frequency. At 68kHz, the current flow is actually less than at 56kHz. It's only 0.806A at 68kHz.
I no longer believe it's the operating frequency that's causing the high current draw.
While it's entirely possible that the replacement FETs are causing the problem, I'd suspect something else.
razirafi:
Have you tried measuring the voltage across the emitter resistors to determine if the problem is excessive current flow through the audio output transistors? |
|
|
| razirafi |
The following three waveforms were taken using my other scope...
The first one is from each bank of push-pull MosFETs on the Gate terminal..
They have the same amplitude.
Vertical: 5V/div
Timebase: 5uSec/div |
|
|
| razirafi |
| This one shows the waveform of one bank of parallel MosFETs on the Gate terminal. The other bank of parallel MosFETs also have the same waveform. |
|
|
| razirafi |
This one shows the Gate waveform on top and the toroid center-tap (+12V supply) on the bottom.
Is this ripple the cause of the overheating filter caps? By the way, is it safe to cut-off the supply rails from the rest of the circuit to check for idle current without any load whatsoever? |
|
|
| Perry Babin |
Did you check the voltage across the emitter resistors to see if one channel was passing too much current?
The ripple is excessive (assuming that the vertical amplifier is set the same as the top waveform). It looks like both banks of FETs are on at the same time causing the ringing.
You can pull the rectifiers to break the connection to the rest of the amp. I've never had a problem doing this but you need to be careful and extra observant in case there is a problem. |
|
|
| razirafi |
Yes, the vertical amplifier settings were the same.
By emitter resistors do you mean those of the output complementary-symmetry transistor stage? They have a value of 0.1 Ohm/2 to 5 Watts each. |
|
|
| Bertje |
These are the emitter resistors indeed.
Or disconnect the power supply from the amp (rectifiers) and put some large resistors on it to make a 5A load. If you measure a symmetrical +/- 30V (.....) the PS is OK. And the amp has a failure....
@Perry: thanks!
Rob. |
|
|
| razirafi |
Hello again guys. I set this thing aside to work on a ZAPCO Z300C2 which would intermittently go into protection mode.
Anyway, I ordered a new 3524 and when I installed it into the Soundstream's circuit, the meltdown scenario was gone:cool:
As for the original 3524 that came with the amp, it promptly found its way into the garbage bin..hehe
Thanks for the interaction guys! |
|
|
| junglejuice |
I see that you have a schematic for this Soundstream amp, can you email it to me? My email is junglejuice72@hotmail.com
Cheers, Junglejuice,,,, |
|
|
|
