• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Pioneer SX-3900 - Idle current too high

princever

princever

AK Subscriber
Subscriber
I picked up a SX-3900 a while ago for peanuts knowing it had tons of problems with the amp board and power supply. I rebuilt the power supply with all new caps and transistors (luckily the 2sk34's were good) and all voltages are within spec. Then I started working on the amp board. I replaced all the transistors and diodes on there, and I had to replace the output transistors with MJL3281/MJL1302 and I made nice thick hold down bars. I used recommendations from the latest MTF list. My board is the later revision type which is the same used in the D7000, so I had extra diodes D135 and D136 which are 4.7V Zener, which I replaced with 1N5230B from Mouser. So I was extra careful and left the output transistors disconnected and fired up on DBT and got my 2nd relay click. I measured the voltages at the connectors and got numbers comparable to the SolutionRoom thread so I connected the outputs and fired up on 150W DBT. It dimmed down ok (117V at receiver) and 2nd relay click again. Measured DC offset and it was high but I was able to adjust using VR1 and VR2 to within a few mv of zero and it stayed there rock solid. So I moved on to the idle current. With pots to minimum I removed the DBT and put it on mains.

Now heres where the problem starts. I have idle current at the test points but the pots (VR3-6) are already at the minimum resistance. The left channel started around 45mv when I first turned it on and slowly increased to 78mv where it stopped. The right channel started around 63mv and slowly increased to 90mv where it stopped after about 15 minutes. I left it running for 2 hours and it didnt go any higher. It is rock solid, doesnt fluctuate around or anything. The heatsink is warm but it doesnt seem any warmer than others of this design type. I nudged the pots a little and it does go up but I cannot get it any lower. Is this a function of the on-semi outputs or should I be able to set it at 56/70mv?
 
Register to hide this ad
I used recommendations from the latest MTF list.
Click to expand...

too many lists floating around to make that a definitive statement.

what are the part numbers of the transistors you used for the on board to-220 heatsink mounted driver transistors?
edit q133 - q136?
 
Last edited:
The Q133/134 originally 2sc2275A replaced with 2sc5171
Q135/136 originally 2sa985 replaced with 2sa1930

The original ones were split (exploded) into multiple pieces. I found many shorted diodes, open resistors, etc. I went through the board and tested every single component. Q101-106 were ok and stayed, all other transistors were replaced even if they tested ok.
 
I also had bad emitter resistors on one channel. I was waiting on MCM for them to come in stock (0.47 ohm 5W) with the dual emitter kind but i got tired of waiting (3 months) so I ordered some wirewound individual 5W 0.47ohm ones from Mouser and joined them for my purpose. I had some concern that these may have some inductance, but I didn't think it would cause this type of issue.
 
I made an umbilical so I could take readings on the amp board with it outside the receiver. The problem I am trying to trace back into the circuit is that the readings at the base of Q133/134 are not symmetrical with that at Q135/136. At Q133 and Q134 I have the requisite 1.2. 1.201 to be exact. At the bases of Q135 and Q136 I have -1.113 and -1.137 to be exact. If my understanding of the drive/output circuit is correct, when these readings are symmetrical the idle current should be zero?
 
Those drivers are the correct choice. Being toshiba, their Vbe shouldn't be an issue.

just for grins, swap the emitter resistors to the other channel, if it continues, they are conclusively exonerated. A pita I admit, but the exoneration will free up a great deal of concern.

princever said:
I made an umbilical so I could take readings on the amp board with it outside the receiver. The problem I am trying to trace back into the circuit is that the readings at the base of Q133/134 are not symmetrical with that at Q135/136. At Q133 and Q134 I have the requisite 1.2. 1.201 to be exact. At the bases of Q135 and Q136 I have -1.113 and -1.137 to be exact. If my understanding of the drive/output circuit is correct, when these readings are symmetrical the idle current should be zero?
Click to expand...

we don't want ANY signal complicating the task of setting the idle current, and the idle current then is just the voltage differential of the combined Vbe turn-on thresholds of the output transistors. and in this situation you want to compare channel to channel.

measure the voltage across R127, and R128, then measure their resistance carefully, and calculate the current, to around 7 milliamps if correct. This current sets the voltage drop across the stv-2 diodes. If that current is reduced, the diode's voltage drop is also reduced. Q115 and Q116 are the current source circuits. If you had a pair of 270 ohm resistors you could try them, for a current of (1.6v / 270 ohms) about 6 milliamps. That should lower the idle current. as a test.

The puzzler is that I have used those transistors in exactly the configuration you describe, successfully, several times - no stv-2 diode current tweaking... It DID have too high of an idle current when I used (incorrectly) mje15032,33's (too slow (30mhz) but at the time I didn't know about the 2sa1930/2sc5171 at 180mhz) and subbed in Fairchild parts (too slow by far at 4 mhz, but as a static test they were ok...)

One channel doing it would be a possible fault, but the chances of a double fault are far smaller. A double error IS possible but so far that is the last thing in my mind...

One thought DOES occur, the power supplies are common to both channels, please check/post the regulated voltages.

measure the operating Vbe of each of the output transistors please. Although it would be hard to tell the small millivolts difference involved. Bet bet is that pair of 270 ohm resistors and the associated STV-2 current/voltage drop.

ps. the "umbilical", good thinking :D, that particular suggestion hasn't been well publicized
(I mentioned a long while back I draw a circuit diagram on cardboard with the wires from the test points appropriately positioned)
 
Last edited:
0. Thanks Mark for helping me. Normally I can troubleshoot through most amps, especially if I have a good channel to check myself, but this circuit design is baffling and extremely complicated.

1. As far as the emitter resistors, I swapped back in a good set of the original Noble dual emitter resistors. One channels was damaged, the other tested ok. So I used those. This did not make any difference in the idle current on the channel I tried. It was still too high and it was the same value as before (90mV).

2. I checked the current through R127 and R128. It seems to creep up quite quickly. here are my measurements. This is early on and it was still climbing but I didn't want to leave it on too long.
R127 - 1.53V/222.7 ohms = 6.8 mA
R128 - 1.59V/220.2 ohms = 7.2 mA

So we are already at 7mA before it even warms up. and this is with both sets of bias pots turned to min resistance. I will try the 270 ohm trick tonight, but isn't doing that really just covering up some underlying problem elsewhere?

3. Voltage measurements. These pin numbers are on the power supply bd
Pin ---( Expected ) --- Actual
22 (54V) 53.51V
23 (32V) 31.1V
24 (-32V) -31.09V
25 (???) -19.9V
26 (-13V) -13.34V
20 (-54V) -53.48V
19 (12.8V) 13.01V
4 (unregulated) 28.89V for protection

I am not sure what the pin 25 is supposed to be, or what the unregulated voltage for the protection circuit is. I think this is ok because I have the newer revision board which uses a positive voltage from my understanding. All the regulated voltages are just slightly low but still within spec.

I will try the 270 ohms for R127 and R128 this evening and see how that goes.
 
Oh, 2 very important voltages I forgot.

The main rails off the filter caps.
+58.7 and -58.7V

The power in my area rarely exceeds 115V. If I had 120 or above I would expect to see the 60V be higher.

So, in summary, tonight I will (First) measure the Vbe for the outputs and (second) try 270 ohm resistors.

Adam
 
Results:

1. The Vbe for all 8 output transistors:

Left
Q1: 0.570V
Q3: 0.572V
Q5: -0.556V
Q7: -0.555V

Right:
Q2: 0.564V
Q4: 0.567V
Q6: -0.550V
Q8: -0.554V

2. The 270 ohm resistors for R127 and R128 aren't enough. The idle current was lower when I started up, but it eventually climbed past 56mv on both channels. I only had it on for 10 minutes so I don't know what the final value would have been when it settled. I can try higher values of resistor if needed.
 
do you still have 4 good OLD outputs of all the wreckage? 2 pnp and 2 npn?

if so it would be instructive to substitute them in, and repeat the idle current adjustment tests, including the Vbe readings.

However:
instead of using the test points which are a weighted summation matrix, start measuring AT/across the emitter resistors, with the current replacement transistors first. That MIGHT illuminate matrix measuring problems, especially if the emitter resistors were damaged (then the matrix resistors, 100 and 200 ohms are straight in the line of fire - ZAPPP) - if that is true (matrix problems) then the idle current measurements are suspect.

an additional observation is that the matrix is one of those circuits that give false reading from paralleled parts, and an open resistor can read ok. THAT's why I say lift a leg of the resistor under test.
 
Last edited:
The emitter resistors are all new because, yes, the originals were damaged. I tried a pair of old ones that tested ok, but that didn't help any, so I put the new ones back in. One channel must have had flaming explosive wreckage because one channels emitter resistors were charred and the 100 & 200 ohm matrix resistors were open and also severely charred. All the 200 ohm and 100 ohm resistors were replaced for both channels. I have verified conductivity around this area of the circuit board where it looks to have gotten crispy.

I do have enough working old outputs to do one channel, but no working driver transistors from the old setup. However, the old outputs were not the correct ones for this amp. They were the 2sa1095 and 2sc2565. Its the same MT-200 package, but not the right part. Should I try them anyway?

I also concern about some of the zener diodes used in this section of the amp. D105/106 is a 6.1V Zener and brand new but only showing 2.45V across it. D135/136 is a 4.7V Zener with about 0.5V across it. Normally when I encounter a Zener in a circuit I can easily tell if its doing its job by measuring across it in circuit. Does this indicate a problem in this area of the circuit?
 
Last edited:
The only advice I can give about the zeners, is to lift a leg, get a power supply, calculate the difference in voltage, divide by 0.007 to get a resistance value, hang a voltmeter across the diode, then use the resistor to supply current to the zener and see where the voltage ends up at.

don't try the non-originals - something DID blow up, more than once...

BTW, the d101- d106 zeners are overload protection, as well as iiac d135, d136
they won't conduct appreciably until the knee voltage.

also don't assume the damage was limited to that channel, with the voltages involved, damage crosses over to the other channel using the protection system as a bridge.
 
Well, yes, I do believe you are correct. I think the reason for the original replacements which were done by someone else was that the connectors for the output transistors were melted and deformed. But they never addressed the connectors which is why it blew again IMHO. But rest assured, I have the new phoenix style connectors in there now and everything is secure.

BTW, thats just my theory on the history of this thing. It's a little like detective work when you try and figure out why someone did something or what happened in the past with one of these.

I tested those zeners (which are new) out of circuit with my bench supply and they are fine. I see what you mean about overload and how they may not reach that voltage at this point.
 
Mark,

I am trying to understand the design of this amp and the one thing I am having trouble understanding is why the matrix resistors are different values. It seems the outputs are almost paralleled, with the exception of their own emitter resistors (as expected). But then we have the 200 ohm and 100 ohm matrix resistors tied together and we take our reading in the middle of them. It also places 300 ohms between the emitters of that respective pair of transistors. Why not use 150 ohms for each? I am not sure how much current is flowing through these during normal operation (although I can check) but it seems that if any current flows it would "weight" the reading of idle current incorrectly, depending on which side is at a higher voltage. If the voltages of the 2 emitters are equal i guess it wouldn't really matter and the reading would be true.

So here is my plan. I am going to take readings across the emitter resistors directly. Like you mentioned before. That should shed some light on the situation.

I still want to know why they are different, 200 and 100 ohms.
 
It is possible the amp is oscillating. If you have an oscilloscope, have a look at the output with no signal. It should be a flat line. If it is oscillating, it will cause high idle current.
 
Well, I scoped the output and it looks okay for the most part, but if I zoom way down to the 10mv/div i see what looks like some noise, a waveform that is around 5mv amplitude with a period of 10ns or so. Its only happening after the protection relay kicks in so I know its the output of the amp. Same on both channels. Not sure if this qualifies as oscillation or just background noise.
 
princever said:
Well, I scoped the output and it looks okay for the most part, but if I zoom way down to the 10mv/div i see what looks like some noise, a waveform that is around 5mv amplitude with a period of 10ns or so. Its only happening after the protection relay kicks in so I know its the output of the amp. Same on both channels. Not sure if this qualifies as oscillation or just background noise.
Click to expand...

Not oscillation. If it was 100mV P-P or more, I'd be concerned. Sounds like the bias circuit needs to be tweaked for the new outputs, they may have higher Beta. I'll have to get the schematic and have a look.
 
That may be the case.

I took some measurements today directly across the emitter resistors.

Left:
78.4mv across both (dual) emitters for that channel
39.2mv across each individual emitter resistor
0.0mv across the 200 and 100 ohm resistors. (sometimes jumped briefly to 0.1mv but usually just sat at 0.0mv

Right:
89.5mv across each dual-emitter resistor
44.7mv across each individual emitter resistor
0.1mv across the 200 and 100 ohm (sometimes saw 0.2mv also)

So to me it looks like things are fairly well balanced with the emitter resistors.
 
I'd like to hear from anyone who has done the bias adjustment on a SX-3900 or SX-D7000. When you turned the bias pots down at the beginning of the procedure, did your the voltage at your measuring points actually go down to zero, or was it just some low number? I would like to know what I am aiming for.
 
I ran across this really old thread about bias being a little high and how AKer DVNOBLE from Brazil fixed it.

http://www.audiokarma.org/forums/showpost.php?p=4033174&postcount=11

I will admit, I was very doubtful, but I did try it. on DBT on one channel with both pots minimized it starts up around 10mv and climbs to around 20. I can then properly do the adjustment. I minimized it again and took it off DBT and it started around 15 and leveled off around 30. I was then able to do the adjust again. I used a 3.3K 1W resistor. Not even warm after 30 mins on.

I am not finding any other options. I even tried all the old small TO-92 transistors which I could find that tested good, and all the old diodes and got the same value (90mv) I tried KSC940TU and KSC2073TU in the driver spots. I know they are slow, I just wanted to see if it would change anything. It didn't. I tried the old outputs even if they arent correct for this model they are from the same series. Nope still got 90mv.

Maybe I can modify several resistors in the bias circuit, so as to not disturb things too badly but just provide some gentle nudging. For example, R127/128 to 270 ohms and R143/144 to 4.7K 2W or something like that.

Also some questions about TH101/TH102. Both channels LOOK ok but with the scorched emitter resistors these thermistors may have gotten quite hot at some point in their life. Right now that thermistor is reading around 10K ohms at around room temp. That means the combination of TH101 and R145, plus R143 is 8.1K. What should the room-temp resistance of that thermistor be if it is working correctly?
 
You must log in or register to reply here.
Back
Top Bottom