MTX 7801 High Idle Current

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have an MTX 7801 that has high idle current. There is an old post that received no replies in the class D section (http://www.diyaudio.com/forums/class-d/274060-repairing-mtx-ta7801-seeking-assistance.html ) that basically mimics my exact situation. I look through one of the older posts also (http://www.diyaudio.com/forums/class-d/274060-repairing-mtx-ta7801-seeking-assistance.html) but there was no real resolution to this.

I have gone through Perry's notes on this amp, but I have not found my root cause yet. The first post that referenced is almost the exact state my amp is in right now. A little history on the amp. I have had it for a couple years just sitting broken. I got to looking at it 2 weeks ago. FET401 (Low side optocoupler regulator) and the output transistor next to it on the heatsink were obviously blown. I replaced those along with a bunch of other found bad parts. I replaced both optocouplers, both opamps in the class D driver stage, one zener around the optocoupler, and 4 different shorted ES2D diodes around the optocouplers, and 2 1000uF electrolytics in the same section. I can power the amp up now and it appears everything is at least turning on. I have the proper +/- 15V on all the opamps (preamp card included). With either the outputs fully removed or the optocouplers removed (with pins 5,6,7 shorted together), I get about 2amps of draw. If I install 1 output FET on each high side/low side, I get about 4.5 amps of draw. If I have all the output FET's installed, same situation. Using an IR camera, the servo NE5532 opamp (closest to optocoupler) shows 90 deg C temperatures with the FET's removed, but goes up to >130C if any amount of FET's are installed.

I have tried a couple other things around the +/- 15V regulators to see if that is the issue, but everything looks solid there. I am looking for suggestions on where to look next. The only thing I have left to check out is the actual output inductors. During the high idle state, they do get warm to the touch and the IR camera shows them at around 60 deg C. Just like the first post I referenced with no replies, the picture there of the oscope output across the speaker terminals is what I have also. I am going to check the output inductors because I would assume that the PWM frequency should not make it through the inductors if they are working properly. They do not look obviously damaged. Any other ideas on where to look besides these inductors?
 
I don't remember the 5532 getting nearly that hot. Could it be defective?

4.5 amps doesn't seem excessive but I don't remember measuring the idle current on one of these.Are the outputs heating up?

The inductors always run hot.

The wires break off of the inductors where they go into the board. Hit the solder connections on them to see if any wires fall free when the solder melts.
 
I don't remember the 5532 getting nearly that hot. Could it be defective?

4.5 amps doesn't seem excessive but I don't remember measuring the idle current on one of these.Are the outputs heating up?

The inductors always run hot.

The wires break off of the inductors where they go into the board. Hit the solder connections on them to see if any wires fall free when the solder melts.

1. I have replaced both 5532's already. Both the originals and the replacements run at the same temperature. I replaced the original NE5532 with the higher temp range SA5532's also.

2. I don't notice the outputs heating up much under the IR camera. The case of the outputs I can scan (the top) stay under 35 or 40C if I remember right. The rest of it is hidden behind the clamps so the IR part always shows cool.

3. I did hit the inductor wires with the iron previously and applied a little more solder. I did not notice anything suspicious when I did this. Unless the adhesive that holds the inductor in place held a broken wire, I did not notice anything abnormal. Should I try to remove all of the original adhesive and do the solder test again? I can always put more goop/RTV on afterwards. Would you expect to see any part of a triangle wave at the (unloaded) outputs if the inductors were intact? Lastly, is there any harm in running the amp with the output inductors completely removed? I would do this test with unloaded speaker output.
 
Last edited:
If the wires are broken, they will generally fall over when you heat the solder. If I'm not mistaken, the inductors are in series. If a wire is broken, you probably would not have continuity between the output transistors and the speaker terminals.

I think the feedback is taken after the inductors so it's unlikely the outputs will oscillate rail to rail with the inductors out of the circuit. This isn't definitive.
 
I checked them again and they seem OK. I fully removed them from the circuit and checked them while attached to the little daughterboard. There was continuity side to side, but not wire A to wire B. This seems to be normal to me. I did not get a chance to power it back up tonight, but I did snap a quick picture for reference. I can't say for sure, but the series 10 ohm and capacitor might be the feedback you were talking about. If it is, then the feedback is indeed BEFORE the inductors and they are in series.

IMG_20180319_190951.jpg
 
I pulled the inductors just because I had not tried that yet and I could at least look at them a little closer. Nothing abnormal with them. They act as shorts side to side and open from opposite sides. The wires all look good and seem in tact. I did power up the amp without the inductors. The amp draws 2 amps (4.5 with inductors installed) in this configuration. I decided to take a couple IR camera scans that I will include below.

The NE5532 near the optocouplers still runs hot. I compared the screenshots of the NE5532 and 3120's in your tutorial versus what I had. Two questions I had on those. First, are those screenshots taken with an audio signal going into the amp or not? I had a 2Vpp 100Hz going into the RCA jacks with the two jumpers installed to bypass the preamp board. Second, is the horizontal scale on your scope pictures 10 microseconds or 10 milliseconds? It looks like microseconds to me, but my particular amp did not look anything like that at 10 microseconds. At 10 milliseconds, they looked identical at the optocouplers. The waveforms at the opamps themselves are quite different though. On the opamp closest to the optocouplers, pin 1 was 0V on the scope. All of the triangle type waves you had were square waves on mine. I don't know if this is due to the fact the inductors were out or not. I will reinstall them and remeasure all the waveforms later.

Op amps near optocouplers
FLIR0620.jpg


Zener near the right optocoupler

FLIR0624.jpg


R117 in the -15V regulator. This seems to be as designed, but runs hot due to it being the full rail to ground through a 10k resistor. That is over 1/2 watt of dissipation.

FLIR0622.jpg
 
This is a self-oscillating amp. With the servo/feedback loop open (inductors out, in this instance), the amp will not oscillate at the normal carrier frequency. All you're likely to get is the signal you drove into it in square wave form. If all of the square waves are square down to the negative supply (pin 5) for the 3120s, they're likely OK.

The flir can be misleading. Is that 5532 so hot that you can in no way hold your finger on it? Not even for a second? I'm assuming that you're not a professional welder that can pick up nearly red-hot work pieces without gloves.
 
This is a self-oscillating amp. With the servo/feedback loop open (inductors out, in this instance), the amp will not oscillate at the normal carrier frequency. All you're likely to get is the signal you drove into it in square wave form. If all of the square waves are square down to the negative supply (pin 5) for the 3120s, they're likely OK.

The flir can be misleading. Is that 5532 so hot that you can in no way hold your finger on it? Not even for a second? I'm assuming that you're not a professional welder that can pick up nearly red-hot work pieces without gloves.

Good to know. I will reinstall the inductors and test the opamp pins again. That would make me think the pictures you have actually are in microseconds, not milliseconds :)

I know FLIR can be misleading. I've done most of the tricks to try to make it as accurate at possible. The opamps themselves don't require too much more to change the emissivity because the plastic black body generally does a good job. Either way, the one opamp certainly does get hot. With the inductors out (or the optocouplers outs), it gets too hot to hold after about 3-4 seconds. When everything is back in, it was one of those situations where you think the part is cool, then you instantly pull your finger back a second later with what you think are second degree burns. When I get everything back in, I will verify the opamp waveforms. If those are good, I'll try to play some actual audio through it. I'm still concerned about that opamp though. I know using originals is always preferred, but would you have any suggestions for an experimental replacement? Maybe something with FET inputs since and lower supply current? The same situation I'm running into has been reported by at least 2 other user threads on the forums here.
 
The reason that it's running hot is because of the load it's driving and the supply voltage. There is a fair amount of power dissipation internally.

I don't have any suggestions for a sub.

If you open the graphic in a new window and right-click to zoom in, you can easily read the scope settings on the lower part of the display.
 
The reason that it's running hot is because of the load it's driving and the supply voltage. There is a fair amount of power dissipation internally.

I don't have any suggestions for a sub.

If you open the graphic in a new window and right-click to zoom in, you can easily read the scope settings on the lower part of the display.

I am going to see what I have in my stash. I did a quick check on the TI website and if I can't find something to try I think I will at least try the OPA1662. Bipolar inputs like the NE5532, slightly higher slew rate, slightly higher gain bandwidth product, half the supply current (so half the power dissipation in the supply case), supports enough current to drive the LED of the optocouplers, and is not very expensive. If I go that route I'll report back, but I will see if I can get the dang thing working in the first place.
 
I reinstalled the output inductors and took some pictures of the resulting waveforms. This is with only 1 output installed per side. Power supply current draw was back up to about 4.5amps. The low side optocoupler supply and the -15V regulator were running warm clamped to the heatsink, but not terribly warm (~65C). The waveforms I have don't look great, but they look sorta close. Sorry about the bad scope, my good one is in use in other location. The opamp closest to the opto's still gets scorching hot. The output inductors were not touchable after 5 minutes at 90C also. But I did get some pictures of waveforms. Included below.

I should have actually done some math before talking about opamp replacements yesterday. While the quiescent current of the opamps is a factor, the load it is driving is much more of a contributor. Something in this design seems like it is off. The LED within the optocoupler seems to be driven WAY harder than the datasheet suggests it should be. In the 3120 datasheet the recommended forward current is 7-16mA with an average absolute max of 25mA. Using the waveforms you have, the opamp puts out 7V and the anode of the opto LED is about 1.5V. That means the 100 ohm resistor has 55mA of current going through it (5.5V/100ohm). I know each optocoupler is only on for 50% of the time, but that is still basically right at/slightly above the absolute average max and a lot above the recommended operating current. Increasing the 100 ohm resistor to something around 350 ohm would make the current live in the recommended range and decrease the power dissipation in the opamp by over a factor of 3. With my opamp output waveforms, I still am over the recommended current (3.5V/100ohm) at 35mA. I am not going to change that yet, but it is something that is in the back of my mind.

All of these pictures are the opamp closest to the optocouplers.

Pin 8 (15V rail). Pin 4 looks identical, but down at -15V

IMG_20180321_183818.jpg


Pin 7 (output to optocouplers before 100 ohm resistor)

IMG_20180321_183743.jpg


Pin 3 of left optocoupler

IMG_20180321_184027.jpg


Pin 1 (output to pin 5 of same opamp)

IMG_20180321_183615.jpg


Pins 2 and 3

IMG_20180321_183719.jpg


I'm not sure what to make out of the big noise spikes on almost everything. It appears on everything that has some attachment to either -15V or +15V. The scope probes were grounded to the chassis ground (which is the same as speaker - on this amp).
 
Generally, if an IC (resistor, whatever) is running that hot, it will affect the solder (it will oxidize and look 'dry'. I've never seen that and I've never known that IC to run that hot (although I can't say I've touched one after the amp has been on a while).

If this is normal, I wouldn't modify it because thousands of amps have been reliable.

If you are going to mod it, I'd recommend getting it working perfectly with the original value components and THEN start subbing parts. Even if the reliability remains the same, you will likely learn something through experimentation.
 
I got the updated pictures. I found a better scope that had better hardcopy features. It certainly improved things using a more local ground. Tell me if everything looks OK. There is still a little noise on it, but not bad at all

Pin 4 of both opamps (-15V)

TEK00000.png


Pin 8 of both opamps (+15V)

TEK00001.png


Pin 1 top op-amp

TEK00002.png


Pin 2 top op-amp

TEK00003.png


Pin 3 top op-amp

TEK00004.png


Pin 5 top op-amp

TEK00005.png


Pin 6 top op-amp

TEK00006.png


Pin 7 top op-amp

TEK00007.png


Pin 3 left optocoupler and pin 2 right optocoupler

TEK00008.png


Pin 5 of right optocoupler

TEK00009.png


Pins 6 and 7 of right optocoupler

TEK00010.png


Pin 8 of right optocoupler

TEK00011.png


Pin 5 left optocoupler

TEK00012.png


Pins 6 and 7 left optocoupler

TEK00013.png


Pin 8 left optocoupler

TEK00014.png


Speaker + terminal with scope probe ground on speaker - (No input signal)

TEK00015.png


Speaker + terminal with scope probe ground on speaker - (2V peak to peak 100hz input signal)

TEK00016.png


Speaker + terminal with scope probe ground on speaker - (2V peak to peak 100hz input signal, zoomed in timewise to show triangle wave still riding sine)

TEK00017.png


So everything looks pretty good except the triangle wave that is riding on the output. Is that normal or does it indicate a problem with the output filter? My inputs on pins 5 and 6 don't look exactly the same either, but close.
 
I've been trying to borrow a working 7801 but no one I know seems to have one. I think I may have an older version that the only real difference in the 5532 drive circuit is a 150 ohm resistor between the 5532 and the optos (which I can change to a 100 ohm). That will let me check the heating if nothing else.

I wouldn't worry about that tiny bit of residual carrier.
 
I've been trying to borrow a working 7801 but no one I know seems to have one. I think I may have an older version that the only real difference in the 5532 drive circuit is a 150 ohm resistor between the 5532 and the optos (which I can change to a 100 ohm). That will let me check the heating if nothing else.

I wouldn't worry about that tiny bit of residual carrier.

Thanks. I'll ignore the residual carrier for now.

As an experiment, I did put a 200 ohm resistor in for the 100 ohm. What ended up happening was that the output voltage jumped up at the output of the opamp. The voltage at the optocouplers stayed the same (which makes sense, it is a "fixed" LED voltage drop). I think with a 200 ohm resistor I had about +/-10V on pin 7 of the opamp. That roughly corresponds to the same current through the resistor with the 100 ohm. It would be even more power out of the opamp though as the output voltage was higher. I have not been able to digest it all yet, but I plan to do more experimentation with this. My first thought is that the design is hitting the opamp output current limit. Based on the schematic you have on the class D driver, there is no reason that changing the output resistor should change the output voltage. There is no feedback from output to input. I am not changing any of the components on the inputs, so to me that further emphasizes my idea at least in my mind. Pins 5/6/7 of that opamp seem to be acting like a comparator for the class D driver, so it should be trying to swing approximately rail to rail at its output. The fact that it does not could indicate it is hitting the output current limit. I will experiment with this later, but I would expect as I keep raising that output resistor value, the output voltage will also keep going up until it hits the opamp's max voltage swing. After that point, as I keep raising the resistor value it will finally start to reduce the current going into the optocouplers. The dissipation could still be higher though because of the increased output voltage. The one downside I see is the output resistor is going to dissipating more power. I have a lot easier time handling that with stacking resistors versus the opamp being scorching hot. It might come down to try actually dropping the resistor value because if it is hitting the current limit, lowering the resistor could keep the current the same, but the output voltage drops so the power dissipation is lower. Couple ideas to check at least.

I will experiment with this later and report back. My idea at least makes sense in my mind, but the best plans usually do not survive the first encounter with reality :) I appreciate you trying to find one to figure out if this is actually normal or not.
 
Last edited:
The 5532 in the old amp with the 150 ohm resistor ran about 55c with a contact type thermometer, so a bit hotter.

Out of curiosity, I built a source-follower pair out of an IRF540 and IRF9540. They were powered off of the same supply as the 5532. I used a 250 ohm 1w resistor on the source to the board. I used a 1k resistor between the 5532, pin 7 and the 540/9540 gates. The 5532 doesn't get hot (only warm). The FETs only very slightly warm up and the 1w resistor is only slightly warm. This amp is on its last FET (literally, 3 of the 4 outputs in one bank are missing) so I couldn't run it hard but it seems OK with the follower in the circuit.
 
The 5532 in the old amp with the 150 ohm resistor ran about 55c with a contact type thermometer, so a bit hotter.

Out of curiosity, I built a source-follower pair out of an IRF540 and IRF9540. They were powered off of the same supply as the 5532. I used a 250 ohm 1w resistor on the source to the board. I used a 1k resistor between the 5532, pin 7 and the 540/9540 gates. The 5532 doesn't get hot (only warm). The FETs only very slightly warm up and the 1w resistor is only slightly warm. This amp is on its last FET (literally, 3 of the 4 outputs in one bank are missing) so I couldn't run it hard but it seems OK with the follower in the circuit.

Thanks for the info and for finding an amp to check this out on. 55C case temp is warm, but not overly so. The FLIR's work a little different that the non-contact ones so I don't know if I can correlate that directly.

In regards to the source follower pair using the 5532 as the gate driver, I'm not sure if I completely understand the similarities to what the 7801 does. Correct me with what I do not understand. It sounds like you are using the 5532 as the gate driver for 540/9540 pair. That will certainly stress the opamp somewhat during the edge transitions, but the "steady state" part of the pulse will be essentially 0 current draw looking into a FET. The 7801's biggest issue I would think for that is the 5532 has to source or sink (depending which opto coupler is being turned on) a fair bit of current during the "steady state" portions of the pulse (~35 to 40mA by my calculation) to turn on the LED inside the optocoupler. I am thinking this is the main source of the heating of the opamp. I might be missing the key point of your experiment, so please correct me.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.