Replacing The STV-3H and -4H Diodes

The stuff I bought at Home Depot had two nozzles that came out next to each other. I think you still have to mix it thoroughly though.
 
They usually make the epoxy resin and the hardener two different colors, that you mix until there are no stripes.
I got these two STV-4H units yesterday, and they both have different numbers from what I have; the originals were both 82A, the replacements are marked 82B and 73C. Should this be a concern, or should I just use the one 82B with the good 82A on the other channel?
 

Attachments

  • 20160116_121903.jpg
    20160116_121903.jpg
    47.8 KB · Views: 393
I would be more concerned with the performance of your diodes than the code stamped on the case. The diagram & processes in post #59 can be used to measure your diodes, but I don't know what a "good reading" would be for the STV-4Hs.
 
If you have a DMM with a diode range you can use that to test them if the meter puts out high enough voltage. If you have a second meter you can measure the voltage on the diode testing meter leads. If it's 2V or more you can test STVs with it.

My Fluke 77II put out 7.3V, can I test the STVs directly using this meter. I'm asking because I have a bad STV-4H, and I was looking at my stash of STVs, All STV-3hs and was measuring them, and getting approx 1.9v across them, I was thinking of just adding a diode in series with it and calling it good. I then remembered a QX-949 parts unit that I have and pulled the two STV-4hs out of it and they measured the same, 1.9v. I thought I would read approx 2.4 volts across the 4-Hs.
 
My Fluke 77II put out 7.3V, can I test the STVs directly using this meter. I'm asking because I have a bad STV-4H, and I was looking at my stash of STVs, All STV-3hs and was measuring them, and getting approx 1.9v across them, I was thinking of just adding a diode in series with it and calling it good. I then remembered a QX-949 parts unit that I have and pulled the two STV-4hs out of it and they measured the same, 1.9v. I thought I would read approx 2.4 volts across the 4-Hs.
It sounds like the current limiting resistor in the meter is dropping too much voltage. You will have to go with plan b. Use a bench power supply or a 9V battery and a 1K resistor in series with it, and connect it to the STV out of circuit. Then measure the voltage drop across the STV.
 
All of this doesn't mean shit unless the diodes track the B-e intrinsic drop and the impedance of the drive/output set that comes after it.

If nothing else, no mater what the voltage drop, if it doesn't track and reduce that voltage enough after you "get the Led out" it could go into thermal runaway.

Bottom line is the voltage drop must drop with temperature FASTER than the output/driver set. This requires not only thrermal coupling, but diode that are NOT advertised as being thermal stable. You want them as thermally unstable as possible. You can always cut back on the compensation in the design, but you will have a hard time upping it though it can be done.

Personally, if I ever design MY amp, I won't use diodes like that, nor will I use that circuit where if the bias pot goes open in commits Hari Kari. I know there is a cost but I have already figured it out, just not for a specific set of parts and power level. My circuit will go to class B if the pot opens. I have seen very few amps designed this way. Very few.

However you are not going to modify everything like that. It is impractical.

Bottom line, the more diodes you string up the more compensation. You normally will have four junctions to compensate, some amps six. You are USUALLY safe if you use that many diodes, and in fact should use one more just for the hell of it and maybe omit the pot altogether.

People who get in this anal set the bias to factory specs are not entirely right. Of course you don't want to bias it too much, and you don't want it in class B, but really if the amp is built right it is not super critical. You are going to crank it into 3 ohm speakers at 200 watts and you are worried about if it dissipates 5 or 10 watts at idle ? Not logical.

The important part is the thermal tracking, period. you need that bias cut down faster than it gets hot. It can be overbiased when cold, a proper design will just cut it down once warmed up. It MUST do that.

Many designs have taken to using a transistor, which amplifies the diode drop. Either the trasnsistor or the diodes are thermally coupled to the heatsink. No matter what, that voltage drop must drop with temperature faster than that of all the outputs and drivers put together. It cannot be at the same rate because the gain of the transistors increase.

Detecting the current does not work because once you crank it up the bias is gone unless you do a hell of alot of engineering. Speaker loads are all over the place, so you can't just design it for an eight ohm resistive load. All that is just barking up the wrong tree because it is a recipe for disaster. Worse than Sony or Mitsubishi ever built, and they did alot worse than the VFET amps, believe me. But if you were to increase bias with more input, that is asking for trouble.

Actually, old, early solid state care radios actually did that, but it was a single ended amp running into a choke. It wasn't really dependent on signal, but the position of the volume control. Even today, most hearing aids are like that, single ended straight into the transducer. When turned down it saves battery power, and the whole design reduces component count and actually is not all that inefficient. And component count is efficiency because people want these things in their ears and small enough they are hard to notice.

But in a push pull amp the bias works against something like 0.44 ohms, not 8. This is what makes it so critical. As far as I am concerned if I cannot get the original bias components for an amp I will run it in class B into a low low load until it is so hot you can barely touch it and then set the bias.

Even then, if it does not track, you run the risk of thermal runaway.

I would consider thermisters. the way most amps are designed we would be talking an NTC, but I would do it the other way. Use PTC. In my design, ANY bad connection causes the bias to stop. It will still work but in class B. I figured out a few implementations of that, one day I might try it. But like Bob Segar said, if you'll sing with me a little bit we might just burn down the house.
 
  • Like
Reactions: trs
The transistor bias control works very well. That's what most amps use. You put the bias pot in the bottom of the voltage divider on the base of the bias transistor, so if it opens, the bias transistor goes into full conduction and turns off the bias. (Some designers didn't think of that and put the pot in the top of the divider, so when the pot opens the transistor shuts off and the amp gets full bias and blows the outputs.)
 
  • Like
Reactions: trs
The transistor bias control works very well. That's what most amps use. You put the bias pot in the bottom of the voltage divider on the base of the bias transistor, so if it opens, the bias transistor goes into full conduction and turns off the bias. (Some designers didn't think of that and put the pot in the top of the divider, so when the pot opens the transistor shuts off and the amp gets full bias and blows the outputs.)

I hearya. But then the engineer has a totally different set of rules. The would like to stay in business so they do things wrong sometimes. Also have seen plenty of it in other equipment. Almost the only time they think about reliability is when it has a super long factory warranty.
 
I tested that a while back, I took a pair of #24 bare copper wires, and made 2 parallel lines 1/16 of an inch apart, then dribbled a jbweld puddle onto them and let it harden.

Then after it hardened, I tested resistance, and leakage current with up to 30 volts across the wires. Nada. down to the microamps.

I tested it because, when the leads break off flush with the stv-xh body, I dig down to good wire (~1/32"), solder on flexible leads and jbweld it all together.

Send me your busted diodes, and I'll try the same for you, I'm actually getting good at it, and have made a mold.

attachment.php

Mark, what material did you use for your mold? I've got to make up a pair of EW-inspired UF1004-based STV-4H workalikes and I was thinking I'd chisel out a hollow space in a piece of plastic cutting board (high-density polyethylene plastic, I think) to make popping out the finished component easy. What worked for you, though?
 
If your cutting board is about half as thick as the desired thickness of your new STV-4 it could work nicely.

Cut 3 pieces of your board so that you get enough to work with. The base piece should be larger.
Drill 2 or 3 matching holes through the all three pieces and mark the other two as top & middle. Place dowels or short bolts in the holes for alignment pins.
Carefully chisel out the desired shape and size in the top & middle.
On mating sides of top & middle cut channels for the leads of the STV-4H. You can put it all in the top or the middle, or you can be fancy & cut ½ into each piece. Use a marker on the upper surface of the top & middle pieces to indicate which lead will be the cathode.
Spray all surfaces that may come in contact with the JB Weld with PAM or a mold release product.
Place the middle piece on the base using the alignment pins and place a small blob of JB weld in the cavity.
Next place your DIY STV-4H in the cavity with the leads in the channels.
Now place the top piece over the alignment pins and clamp the three pieces together.
Ensure that your STV-4H is correctly centered, then fill the rest of the cavity with JB weld.
After the epoxy is dry, mark the cathode lead BEFORE you remove it from the mold.
Separate the three pieces of cutting board and you should be able to easily remove your new STV-4H.

Hope I didn't miss anything.
 
I've taken some measurements of SV03 variants and SV04. They seem to measure quite different to the STV-03H and STV-04H. Maybe relevant to folks stumbling onto this thread. A work in progress for suitable sub... results being posted over in am Sansui AU-X11 thread.
 
I've taken some measurements of SV03 variants and SV04. They seem to measure quite different to the STV-03H and STV-04H. Maybe relevant to folks stumbling onto this thread. A work in progress for suitable sub... results being posted over in am Sansui AU-X11 thread.

Update in this post.

Summary is:
SV03 --> 3x 1N4150
SV04 --> 4x 1N4150.
 
> All of this doesn't mean shit unless the diodes track the B-e intrinsic drop and the impedance of the drive/output set that comes after it.
So true. A diode junction is about 0.7v a small change in in this goes all the way from turn off to saturation. It is the transistor bias voltage at all temperature that is needed to control.

>The transistor bias control works very well. That's what most amps use.
It does work well if the transistor is put on the output heat sink also.

>Almost the only time they think about reliability is when it has a super long factory warranty.
The "they" are the beam counters. Engineers know better but they are not allowed to do it right at times. Right often means cost.
 
Bumping this thread - in replacing a DC offset pot in my Pioneer M22, of course I broke the leads on probably the worst component to do this on - the STV-4H. Reading Hyperions' original post here - is four UF1004's in series still a good replacement? Their dimensions seem to be OK to fit into the amplifier's heatsink recess.

ARGH. Thanks
 
I've found that four UF1004's track the temperature pretty well. But on something like the Pioneer running in Class A, I'd try real hard to locate a real STV-4H diode.

I'm working a deal now with a seller who has a few of them. If it goes through, we'll see about getting you one.
 
Bumping this thread - in replacing a DC offset pot in my Pioneer M22, of course I broke the leads on probably the worst component to do this on - the STV-4H. Reading Hyperions' original post here - is four UF1004's in series still a good replacement? Their dimensions seem to be OK to fit into the amplifier's heatsink recess.

ARGH. Thanks

I've found that four UF1004's track the temperature pretty well. But on something like the Pioneer running in Class A, I'd try real hard to locate a real STV-4H diode.

I'm working a deal now with a seller who has a few of them. If it goes through, we'll see about getting you one.
If Glenn is unable to help, I have a couple STV-4H's and will make one available.
 
I tested that a while back, I took a pair of #24 bare copper wires, and made 2 parallel lines 1/16 of an inch apart, then dribbled a jbweld puddle onto them and let it harden.

Then after it hardened, I tested resistance, and leakage current with up to 30 volts across the wires. Nada. down to the microamps.

I tested it because, when the leads break off flush with the stv-xh body, I dig down to good wire (~1/32"), solder on flexible leads and jbweld it all together.

Send me your busted diodes, and I'll try the same for you, I'm actually getting good at it, and have made a mold.

View attachment 817390
Thanks you very much Mark, I busted one wire off of my daughters 2220b last night while recapping. I had already started exposing the stubbed wire with my dremel when I ran across this thread. I must say, it was nice to know that somebody had successfully tried it before and was able to walk me through the procedure. Holding my breath and crossing my fingers that it will work tomorrow when I reassemble the amp board.:beerchug:
 
Back
Top Bottom