quad silicon bias diode

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The PV1.3k BJT amp has a quad series temperature sense diodes in one axial package on the heat sink. I bought one with the output transistors mostly blown. One of these diodes with 19v through 500 ohms on it reads 2.7 v forward (19 backwards), the other reads 2.2v. I think probably the 2.2 v reading one is damaged. I see in this thread http://music-electronics-forum.com/t20768/ Enzo reccommends visiting my Peavey dealer, which is 3 miles from the bus stop and probably will have to order it anyway, requiring two six mile walks. I can't find anything like this at my usual suppliers, newark & mouser. Instead of two SZ-13886-2 double series silicon diodes in series like the schematic says, the unit has single fat 2 leaded axial components stuck in a hole in the heat sink. Can I use a 2.7v zener diode for this? Does the thermal coefficient go the right way? Or can I use two old red LED's in series with the emmiter part stuck in a hole? I don't want to use 4 each 1n4148 in series, the metal leads are too close to the heat sink and using electrical tape destroys the thermal coupling. If I drill 3 more holes I can use 4 TO92 transistors in series- would the BE junctions be best, or the CB junctions? This thing has 95 V rails, I don't know how much the temp sense diode has to hold off.
Oh, I read wikipedia about silicon diodes, and the zener and 1n4148 datasheet. Nothing about thermal coefficient. ??? Thanks
 
Last edited:
You can't edit here after 1/2 hour. Here is more of the story if anybody is interested in anything this elementary. A Shanghaisun datasheet for 1n5224 zener diodes (2.8v) says the temperature coefficient is <-.08 v/degK. The peavey semi cross sheet posted here on diyaudio has a footnote at the bottom that says a Peavey SQ13886-2 is a specially selected diode from a run of MZ2361. A microsemi datasheet for that parts says it has two diodes in series, and each diode has a temp coefficient of ~-.002 V/degC. So zeners won't work. Peavey drilled an extra hole in the opposing same channel heatsink, and the cool thing about their machine work is that there is a tiny lead sized hole at the bottom through the PCB. So if I insert a 1N4148 over there, and put it series with the 2.2V diode on that side, I will have about 2.8v with all silicon diodes. I'll try that first cut. Possibly Peavey selected this part to set idle current, and I've installed terminal strips so I can measure idle current conveniently from the top. If I have too much idle current I'll use a resistor soldered on the driver pcb to turn it down.
Following the suggestion on the music-electronics forum sort of, I'll put a 3.3V zener across these pins to the output board on the driver board. Then if the tinned connector to the output board corrodes up maybe the idle current won't go to infinity. CH on here fixed a similar design PV8, that had a corroded ground pin from driver to O.T. assy, so I expect it to happen someday.
Oh, and microsemi makes on the same datasheet a MPD400 which is 4 silicon diodes in series. They call it a stabistor. Mouser doesn't stock it either. The 2.2V diode that came out says "DZ9905 22BN05" which crosses to nothing. I'll hope it is supposed to be a 2.2v diode, or that if one diode is shorted, it stays that way. The original settings of the bias are not interesting anyway, I have all new output transistors and they are not MJ15024 and MJ15025 anyway.
 
Last edited:
Well, if you don't want to physically go to a Peavey dealer, you can pick up the phone and call the factory, who will be happy to send you the exact parts they use at reasonable prices. I think that is what I suggested over there, rather than a personal visit.


The previous owner replaced the temp compensation diodes in the bias circuit with something else, and the amp is now blown up? No surprise.

Just get the right part, really.

The amp has two identical channels, so there should be a pair of these diodes on each channel. You mentiion you checkes two, one was 2.2v and the other 2.7v, and you think the 2.2 is the bad one. I'd be thinking the 2.2 was the good one. But compare to the other channel

Are you sure about how the circuit works? You mention 95v rails and these diodes. Look at the schematic. The whole point of the bias string in an amp like this is to maintain a voltage space between the bases of the opposing sides output transistors. It must do this with the help of the drivers and predrivers, so really the bias string spaces the predriver bases apart.

For sake of discussion, let's call a junction drop a half volt, so for discussion, a diode has a .5v drop, and a transistor B-E also has a .5v drop. So considering the output bus ground, then the opposing (V+ side and V- side) output transistor bases would start conducting at +0.5v and -0.5v, which would then be their bias point. Back on the driver card, the drivers, whose emitters connect to the output bases, would then expect to have 1v on their bases, and then the predrivers another half volt from there. SO the predrivers want their bases at 1.5v. And from one side to the other then is a 3v space. No matter what the signal does, that group should always have the same 3v across itself. That is the bias. Real world voltages will differe some, but hopefully the concept is clear.

Now in the PV1.3K, on the driver board you see the circuit as follows: from the base of the V+ predriver Q100, the bias circuit flows through CR105,CR107 (1N4148 with a parallel resistor on one) then through the ribbon over to the two dual diodes CR139,CR140 on the output heatsink, then return and connect to the base of V- predriver Q104. So what you have is six diode drops in series separating those two bases. In my example, six drops is 3v, just exactly what we wanted.

That sets the transistors just on the threshold of conduction. ANy less voltage space and you get crossover distortion, any greater voltage and the opposing sides will both be forward biased and will conduct at the same time, causing the amp to run hot.

So if any of the diodes go high in their drops, the amp will run hotter. In fact, you could short from pin 1 to pin 2 on the ribbon - shorting across the pair of dual diodes - and accept the increase in crossover distortion but would be able to check out the rest of the circuit. In fact one factory recommended fix if the amp seems OK but runs hot is to short across CR105.


To get that string of voltage drops, current must flow through the string. Yes, the amp has the 95v rails or whatever, but look on the driver drawing top center. The V+ rail is fed through 4k 5w R100 to CR100 16v zener. R102 and CR104 feed that 16v to the base of Q100. The same thing happens on the V- side. SO your diode string is running between +/-16v or less. Not 95v, unless a bunch of other parts fail.

I am sure we could engineer some alternative way to do this, but the existing circuit is simple and reliable and cheap. I recommend sticking with it.
 
Thanks for the explanation. No, I really didn't have a clue how this works, just something about voltage space between the two drivers. I read through an explanation of a 2 diode bias circuit in the GE semiconductor handbook 7th edition 1966, which was of course a lot simpler. That didn't cover heat sensing at all. I suppose I should really buy Bob Cordell's book. The fact that this quad diode stack runs off the input op amp (4558) has me completely bamboozled.
I found the other double diode on each side buried in white compound, so I do have 2 double diodes on one side (2.7v) , and a double diode buried in compound and a single diode in the heat sink hole, the 2.2v side. The single diode reads 560 ohm on my meter just like a 1n4148, and the doubles won't read except on the 2 meg ohms scale, so I think the double diodes are okay and the single was put in by a previous tech who wanted to finish the job & ship it.
Thanks for the tip on calling the factory, I might. Other double forward diodes are 1n4156, 1n4830, MPD200, and of course MZ2361. American Microsemiconductor Online Store Home Page manufacturers of Diodes, Transistors, Thyristors, Triacs, Diode Array, Integrated Circuits and Semiconductors -- Online Store has the MPD200 @ $24 each, and the 1n4156 at $6.30 each. Avnet has got 4 triple 1n4157's, but they have a $50 minimum now, pity, they used to be my favorite supplier. Cheapskate that I am I was hoping to combine an order with something else I need (all these dead organs with dried up 40 year old electrolytic caps) but I'll have to order something special I believe. Right now I'm trying to solder two 1n4148 together and heat shrink them and stuff them down the hole in the heat sink, but the glass package + heat shrink is stupidly insulating. Tried a TO220 TIP31 (for a diode) on the heat sink, even drilling a new hole it won't fit anywhere. Thanks again.
 
Last edited:
You should be using the diode test function of the meter instead of the ohms function to check diodes.

The factory schematics include parts layout drawings, if you don;t have the complete file, contact customer service at Peavey for it.

What moves the bias astring around doesn;t matter a lot. the concept here is pretty straightforward. Betwen the split supplies, the outputs are placed, and the drivers and predrivers arranged with the bias string so the bases are just about to turn the transistors on. No matter what, that 3v space is maintained. Assume this is all we have, nothing further left on the schematic exists yet. Since the transistors are not yet conducting, just ready to, the output bus will remain at zero if the bases are centered around zero. (I am speaking generally here, because this grounded emitter configuration can't have anything but zero on the output bus.) Now if I push the bias string up or down, that space remains, but the bases will all move the same direction, either more positive or more negative. AND THE OUTPUT BUS WILL FOLLOW. That is how the amp works. We get something to drag the bias string more positive, the outputs follw it more positive. Get it to go more negative, the outputs follow. And it doesn;t matter where we grab that bias string. move the top diode, the bottom diode, or one in the middle, all of them will move as one.

Those resistors and the +/-16v supplies want to center the string around zero.

Think of the front wheels on your car. There is a tie rod that connects the left and right one. Whichever way you turn one, the other moves the same way. You can grab the left tire and make the right tire move, or you can grab the center of the tie rod and move them both from there. Your string of diodes is that tie rod then. This amp uses a string of diodes, many more amps use a transistor, doesn;t matter, as long as whatever we do maintains that space.

On a more common design, there would be a voltage amplifier stage - a VAS - and it would connect to the bias string, and the music signal would drive the string up and down with the waveform, and the output will follow.

On this amp, instead of a voltage amp transistor, there is an op amp. The op amp is attached to the bias string, so whatever signal comes out the op amp, it will move the string up and down, and the output will follow. Transistor, op amp, steam shovel, doesn;t matter, whatever we come up with to apply our signal to the bias string will also move the output.

The heat sensing is a bonus, but the main job of the diodes here is to maintain the space, the bias. Of course there is a lot more to an amp than this simple conceptualization, but I am not designing them, I am troubleshooting.

In a smaller amp, I might not have predrivers, I might have only outputs and drivers. In stat case, there will be one less junction on each side, so a 2v space instead of three, and that would mean two fewer diodes. My old GE transistor book is 20 miles from here, or I'd look up their circuits.


So you have discovered the "other" diode on each end? Were your 2.2 and 2.7 volt readings across the PAIR of dual diodes then? (Four drops)
 
Yes, the 2.2V and 2.7V readings were from pin 1 to 2 of the connector, so it was across all diodes. Having found the terminals on the back, on the "good" side I'm reading 1.48 and 1.49 v on the two diodes, and on the "bad" side the one buried in the white compound is reading 1.48v. This is a 19V source through a 500 ohm resistor, so test current is about 35 ma. I lost the "single" diode last night. It is colder today, 58 deg F in here, probably why the voltages are higher.
My previous amp I repaired the dynakit ST120 had the bias current (double diode from top rail) applied to the emitter junction of the driver transistors, not the bases where the voltage amp stage drives. So the bias and signal were separated by the driver transistors. That was easier to comprehend. I see now the 4558 drives Q100 the NPN predriver directly, but it drives Q104 the PNP predriver only after the diode stack.
I'll put the thermally insulated 1n4148 stack in for test this weekend, then walk up to Far Out Music Monday and see what they can sell me for the real double diode in a DO41 pack. (Mouser has SOT but it won't fit in the hole in the heat sink, and I can't cut small square holes).
I had hoped that finding CR105 shorted would explain why Q100 Q105 and the NPN O.T.'s are hard on (+65dc @ speaker terminal) and Q104 Q106 and the PNP OT's are off but apparently it has nothing to do with that. The quad diode stack on that side is okay. Drat. I keep hoping Q102 will turn off Q100 due to the 2.2k O.T. emmiter sense resistors, but it won't. I'm getting 0V off pin 5 the feedback, ??? The resistors are not blown and the connector has continuity from the resistors to the Q102 base. Thanks for explaining.
 
Last edited:
An opan diode in the bias string could cause the output to swing over to a rail, or try to swing to both. POOF. But a shorted diode would just take the transistors farther from conduction and the amp would run cooler but with more crossover distortion.

I have been a Peavey service center for 25 years, and in that time I have never bought those dual diodes from anyone but Peavey. I never saw the need to invent something different or find a substitute.

It is alwaqys tempting to look for shorted parts, but don;t overlook that an open resistor can be just as much a problem. Or an open connection between parts.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.