freestompboxes.org

Hi all,

This topic might as well been in the Workbench subforum, but it's all about an amp. I detail the debugging process for further information, but my quetsion will come afterwards!

So a friend of mine brought his faulty RS700 head this morning. Symptoms are:

- the amp buzzes loudly when no instrument is plugged (even with no speaker on).
- one of the two heatsinks (the one for the P-channel mosfets) is getting really hot.
- when he plays the instrument, fuse blows at first note...

When power transistors are unplugged from the power supply, I get the +75/-75VDC ok.

When the power transistors are connected again, the 75VDC wire appears to be 0VDC on the N-channel side. And, of course, the -75V turns into -150VDC for the P-channel side...!!

So there's a short, obviously. Quick check of the 5 resistors, ceramics and mosfets brings out one N-channel with Drain and Source shorted. Part is around 8€, way cheaper than UPSing the bloody unit back to germany

All done!

Now, there are 5 mosfets in parallel, like in this picture below (Ampeg SVT4Pro): Do I need to match the new mosfet with the 4 others, or having these running in parallel is going to "compensate" the possible differences?

What parameter do I have to look at when matching?

Thanks for your help on this!

lolbou wrote:Hi all,

This topic might as well been in the Workbench subforum, but it's all about an amp. I detail the debugging process for further information, but my quetsion will come afterwards!

So a friend of mine brought his faulty RS700 head this morning. Symptoms are:

- the amp buzzes loudly when no instrument is plugged (even with no speaker on).
- one of the two heatsinks (the one for the P-channel mosfets) is getting really hot.
- when he plays the instrument, fuse blows at first note...

When power transistors are unplugged from the power supply, I get the +75/-75VDC ok.

When the power transistors are connected again, the 75VDC wire appears to be 0VDC on the N-channel side. And, of course, the -75V turns into -150VDC for the P-channel side...!!

So there's a short, obviously. Quick check of the 5 resistors, ceramics and mosfets brings out one N-channel with Drain and Source shorted. Part is around 8€, way cheaper than UPSing the bloody unit back to germany

All done!

Now, there are 5 mosfets in parallel, like in this picture below (Ampeg SVT4Pro):

ampeg-power.gif

Do I need to match the new mosfet with the 4 others, or having these running in parallel is going to "compensate" the possible differences?

What parameter do I have to look at when matching?

Thanks for your help on this!

I'm not sure if you do have to match them, but if you do then I guess you probably have to match them for gate pinch-off voltage, and gm (transconductance)...anyway, best that I can think of off the top of my head, Lolbou....

Thanks for your help Simon.

DrNomis wrote:I guess you probably have to match them for gate pinch-off voltage, and gm (transconductance)...

I sorta thought about gm, but I really wonder if it's needed in such a pattern (i.e. paralleled mosfets).

The only advantage I can see in matching these is to equally share the current they have to provide for the 5 N-channel ones, and to balance the signal symmetry with what the P-channels are providing on the other side... But again, will it be that critical?

lolbou wrote:Thanks for your help Simon.

DrNomis wrote:I guess you probably have to match them for gate pinch-off voltage, and gm (transconductance)...

I sorta thought about gm, but I really wonder if it's needed in such a pattern (i.e. paralleled mosfets).

The only advantage I can see in matching these is to equally share the current they have to provide for the 5 N-channel ones, and to balance the signal symmetry with what the P-channels are providing on the other side... But again, will it be that critical?

No worries Lolbou, hope you manage to fix the amp....


Well, I do know that for power amps using Tubes/Valves, you do have to match them up, a Vox AC 30 is a good example, it uses 4X EL84 Tubes, with two in Push-Pull Parallel, the bass amp you're servicing uses devices which are like the Solid-State equivalent of Tubes/Valves, but the only real benefit in using matched devices is probably a lowering of distortion when the amp is operated near clipping, particularly Even-Order Harmonics, I reckon you'll most likely be okay since those devices that the amp uses were manufactured within fairly close tolerances, so there won't be that much difference from device to device, you could try checking the amp out with a scope, signal generator, and a dummy load, to see how it behaves after soldering in the new device, be careful powering the amp up though, something made the original power Mosfet go short-circuit, maybe another unrelated fault?....

Yes. You most definitely have to match them. The matching - in fact replacement and matching of all power transistors - in this kind of repair is advisable even for plain BJT transistors and it's pretty much mandatory for MOSFETs, as they have parameters scattered all around.

In some cases, just buying all of them from same production batch is enough, though. However, I cannot vouch that the trick works in this case.

MOSFETs with positive tempco are more tolerant towards mismatches, MOSFETs with negative tempco need to be matche very, very carefully to avoid current hogging.

AFAIK, you need to measure that Vgs (under proper loading) fall within very close range and that current draw under that very same loading also falls to very close range.

Thanks!

teemuk wrote:pretty much mandatory for MOSFETs, as they have parameters scattered all around

Can you point out which parameters?

teemuk wrote:However, I cannot vouch that the trick works in this case.

You mean because of the circuitry itself?

lolbou wrote:Can you point out which parameters?

Pretty much ALL of them. Those are notoriously very inconsistent devices that have a very large parameter spread.

You mean because of the circuitry itself?

No, because of the MOSFETs themselves. It really depends on how close tolerance the manufacturer can end up with each MOSFET within the manufactured batch. You can google search with the device type and see if folks have experiences of that. It's mostly just information that compiles as reputation along a pretty lenghty time frame. The trick worked with some old Hitachi MOSFETs for example, as each manufacturing batch had a very small parameter spread, but it's not bound to work with each and every MOSFET from who knows how many different manufacturers.

The circuitry by itself already aims to ideal current sharing by using close matched source resistors that try to even out the current draw of each device. However, it can only do a little if one or two MOSFETs happen to have very different parameters than the rest; they will current hog in that case despite the current sharing resistors. Matching is needed to ensure the scheme can actually work.

teemuk wrote:MOSFETs with positive tempco are more tolerant towards mismatches, MOSFETs with negative tempco need to be matche very, very carefully to avoid current hogging.

Sorry, this should have been:
MOSFETs with negative tempco are more tolerant towards mismatches, MOSFETs with positive tempco need to be matched very, very carefully to avoid current hogging.

I may add the part references...

The broken mosfet is 2SK1058 and the complementary are 2SJ162, both made by Renesas. It is said here that they've taken over Hitachi's business...

Actually my Peak DCA can indicate me the Vgs value. I may desolder the four others to have an idea of their value... Unfortunately I don't know how to check for the other parameters.

teemuk wrote:MOSFETs with negative tempco are more tolerant towards mismatches, MOSFETs with positive tempco need to be matched very, very carefully to avoid current hogging.

Shall I look at the Id vs. vgs curve on the datasheet for this "tempco"?

lolbou wrote:I may add the part references...

The broken mosfet is 2SK1058 and the complementary are 2SJ162, both made by Renesas. It is said here that they've taken over Hitachi's business...

Actually my Peak DCA can indicate me the Vgs value. I may desolder the four others to have an idea of their value... Unfortunately I don't know how to check for the other parameters.

teemuk wrote:MOSFETs with negative tempco are more tolerant towards mismatches, MOSFETs with positive tempco need to be matched very, very carefully to avoid current hogging.

Shall I look at the Id vs. vgs curve on the datasheet for this "tempco"?

You could try looking at the Id Vs Vgs curve, but bear in mind that the curve may be for one particular device, and sometimes the curve is for the average of a group of devices.....

Try to match them, but don't worry too much about it with those units.

Just get the overall bias..which is what like -2.5mv or something? and make sure you get 0VDC at the output.

Fucking mosfets for output... I hate'em.

Thans for the input Mike. I've repaired it, simply dropping in the new MOSFET. I did not check the 0VDC at output nor the bias, but just plugged it straight : sounded perfect, even at BIG volume.

chicago_mike wrote:Fucking mosfets for output... I hate'em.

Because of reliability or sound?

Bit of both.

I prefer the Sanken output transistors. 2SC3264 and its pnp counterpart. Great power and sound.

the mosfets amps, eden and some earlier hartke's...always in the shop.

chicago_mike wrote:the mosfets amps, eden and some earlier hartke's...always in the shop.

I can't remember "my" bassist having any issue with its Ampeg SVT4-Pro... These are mosfets powered too, aren't they?

We get those in from time to time. Actually we do have one in now.

With Ampeg..the biggest gripe I have is the cheese ass ribbon connectors. We replace those all the time. They're not really meant to be in a big ass vibrating bass amp.

chicago_mike wrote:We get those in from time to time. Actually we do have one in now.

With Ampeg..the biggest gripe I have is the cheese ass ribbon connectors. We replace those all the time. They're not really meant to be in a big ass vibrating bass amp.

Says Mike and the IDC-riddled Thunderfucker.

What is their most common mode of failure? Wires unseating? I'm considering the TE MT156 range for my amp interconnects.

The connectors on the pcb themselves...cracked solder connections every damn time! and the ribbon cables they use...not flexible enough for a vibrating bass amp. Just horrid design.

Especially the SVT 2 Pro!!!!!! Total fuck shit!!