Replacing The STV-3H and -4H Diodes

For those of you fabricating replacements and attachment methods, allow me to show what I've just discovered under the hood of the third HK 330B variant I've worked on. This is the original, from-the-factory mounting method for the STVwhateverH in this unit. It isn't rocket science or perfect, but it might lead to additional ideas on how to attach these things.

IMG_0309.png IMG_0310.png IMG_0311.png IMG_0312.png
 
Hello. Resurrecting this thread again:

What do you recommend finally to replace STV-4H in a SCOTT 460A? The "3x4148 + jb weld" route, or the "transistor + 2 resistors" circuit posted before ?
I've seen a similar transistor circuit recently mounted on a Harman Kardon HK6500, I could duplicate the set, and tweak the resistor values. The diode route seems easier.

Do you recommend to replace both channels, or just the bad one ? I'm having problems to bias only one channel, and I measured a voltage drop of 2.1V across the diode in that channel, compared to 2.6V in the good one.
 
Hi elnaldo, l don't know if this helps or not but here are two of my threads. For a Sony l used information from a factory Sony bulletin to replace the SV04's, you could possibly change the resistor values to replicate operation of a STV-4H, l like this setup because it is easy to mount to heat sink.
http://audiokarma.org/forums/index.php?threads/sony-ta-2650-bias-diode-replacement.766998/

I used three 1N4007's in series to replicate a STV-3 in a Kenwood but l think you could get pretty close to a STV-4H by using 4 in series.
http://audiokarma.org/forums/index....-bias-adjustment-and-protection-relay.736420/
 
Thanks ! I think I'll try both solutions. I've tried already 3 x 4148 diodes and they work OK (4 diodes was too much voltage drop, too high bias impossible to adjust).

But I see the transistor solution looks better than the epoxi thing. I have different NPN spare transistors to try.
 
Well, this was my workaround with 4 x 4148 diodes. After one try, had to cut the epoxi and jump one diode to leave just 3.

At the end, I didn't use it, since the main problem was at other point. Fixing that, allowed me to adjust the bias with the original STV-4H diode.

P1160038.jpg

P1160040.jpg
 

Attachments

  • P1160039.jpg
    P1160039.jpg
    38.2 KB · Views: 75
Last edited:
Well, this was my workaround with 4 x 4148 diodes. After one try, had to cut the epoxi and jump one diode to leave just 3.

At the end, I didn't use it, since the main problem was at other point. Fixing that, allowed me to adjust the bias with the original STV-4H diode.

That's a very neat solution elnaldo :thumbsup:.
 
I didn't use it in the amplifier. It's this same "component" but one diode jumped inside. So, 3 diodes inside.
 
From help from this Forum - I assembled 3 in series UF1004 Diodes onto a Ring Terminal.
With this Varistor in Free Air dangling.
Operated the Receiver for 2 hours thus far with Good Output Power Amp DC Balance Voltage results, + 8 mV.

Will continue this testing with this UF1004 Ring Terminal mod attachment to power transistor heatsink.
 
To replace the STV-4H's in an SX-5570 that came to me in a rather beat up condition, I fashioned up the attached jury rig using some perfboard, 3xUF4007, thermal compound and the testing methods described in this thread.

The forward voltage in the 10's of mA's is noticably lower than the STV-4H (went down by about 300mv), but temperature response is comparable at those currents.

Initial bulb-limited power-up, full power-up and first bias-setting reveal that it's a fairly stable hack, however, could someone offer some reassurance this can work for a long run-time as well?

EDIT:
For those googling into this thread in order to find a quick and easy STV-4H replacement that doesn't cost a lot of money, time and epoxy:

The perfboard-3xUF4007-perfboard-sandwich shown in the picture proved in my case to be a suitable and stable substitute after a few hours of intensive use, but do make sure there's plenty of thermal contact.

I took the following (highly scientific) measurements in my 5570: When properly adjusted, voltage drop across current sharing resistors is in the <10mV ballpark on cold start, goes up to 20mV after 10 minutes of easy listening, and will drop to 18mV after blasting into 4 ohm load for 10 minutes or until heatsink is toasty. For attaining that proper adjustment, of course reference the service manual.
 

Attachments

  • WP_20171202_002.jpg
    WP_20171202_002.jpg
    61.3 KB · Views: 324
Last edited:
So, it's wire in, three diodes zigzag woven through the perfboard, out from the last one, and the whole assembly secured through the old bolt hole and the diodes in close thermal and physical contact with the heat sink. Elegant.
 
So, it's wire in, three diodes zigzag woven through the perfboard, out from the last one, and the whole assembly secured through the old bolt hole and the diodes in close thermal and physical contact with the heat sink. Elegant.

Thanks! That's it, yeah. I want to give it a good week of playing at moderate levels before I can recommend it definitively. Adding detailed crop of hack later, along with definitive verdict.

I'm still a bit worried about mechanical stress on the diode legs that might occur with temperature changes.
 
My homebrew STV-3's are still working fine on the Kenwood I repaired, haven't had any trouble with the amp at family-member-complaint volume levels, though I've not tortured it particularly. Bias moves around a little bit with heat but the unit has never gone into protection or blown fuses etc.

If mechanical stress is of concern, perhaps add a small rubber o-ring between the nut and board on your binding screw- to allow things to flex as needed.
 
Here's what I did on a Marantz 2230. Worked like a champ.
I haven't had to try this yet - thanks to great members of AK helping me out with getting original parts....!

But for future reference - is it safe to assume that the epoxy these diodes are mounted in will conduct heat nicely and in a similar way to the original part - ie: where the diode replacement jobby is used for thermal heatsink tracking?
 
I haven't had to try this yet - thanks to great members of AK helping me out with getting original parts....!

But for future reference - is it safe to assume that the epoxy these diodes are mounted in will conduct heat nicely and in a similar way to the original part - ie: where the diode replacement jobby is used for thermal heatsink tracking?

The diodes "sank to the bottom" when I poured the epoxy, and that side got placed up against the sink, so the diodes were almost in direct contact with the sink. If I had to do it again, I would arrange the diodes in a more compact and efficient "Z" configuration.

The channel with the improvised epoxy replacement stayed as cool to the touch as the channel with the original. Bias set to the proper value and was stable after hours and hours of up time.

Bias Diode.jpg
 
Last edited:
About 30 hours in, the perfboard solution continues to hold up well.

It's not as pretty as epoxylug can be, but at this point I think I can recommend it. 3xUF4007 seems to be suitable as well.

I'll be trying the epoxylug next time I need a replacement, getting perfboards down to this size without jeopardizing structural integrity is no fun with the logs I have for fingers.
 
Back
Top Bottom