Matched Toshiba Mosfets?
You guys seen this yet?
http://www.diyaudio.com/forums/swap-meet/161366-matched-toshiba-fets-sale.html
These are the exact Toshiba Mosfets Elliot mentions on his site as a 5th alternative to fix the output stages on the SA amps that "some people are using successfully". Models are 2SK 1530 for N-type and 2SJ 201 for P-type.
This member Blues still has matched pairs of N and P types that he is selling for $29 a pair plus shipping. The only thing is that even if I wanted to pick some of these up as a back up plan I wouldn't know which ones to chose based on the measurements taken for Vgs? The only thing I have to go by as far as possible matching measurements on the stock Mosfets is the numbers I found scribbled on the metal cases I saw when I pulled the plastic caps of last night "160" and "162".
You guys seen this yet?
http://www.diyaudio.com/forums/swap-meet/161366-matched-toshiba-fets-sale.html
These are the exact Toshiba Mosfets Elliot mentions on his site as a 5th alternative to fix the output stages on the SA amps that "some people are using successfully". Models are 2SK 1530 for N-type and 2SJ 201 for P-type.
This member Blues still has matched pairs of N and P types that he is selling for $29 a pair plus shipping. The only thing is that even if I wanted to pick some of these up as a back up plan I wouldn't know which ones to chose based on the measurements taken for Vgs? The only thing I have to go by as far as possible matching measurements on the stock Mosfets is the numbers I found scribbled on the metal cases I saw when I pulled the plastic caps of last night "160" and "162".
On the top of his spreadsheet he mentions that the Vgs = 2.000+ and the numbers in his spreadsheet like 0.303 is what you add to get the Vgs (2.303v). The actual Vgs is not so important, it's the matching. Though, he charges more for the higher Vgs values, so maybe they are of higher quality? I don't know. Looks like they are very closely matched just like Elliots.
That could be a good route to take.
Hi,
IRF parts no longer available? That's what was used for the SA-100 and SA-220 imported up here.
One word on device matching. The original parts were not matched that well to begin with. I suspect they were just tight enough to do the job. At any rate, use the IRF parts and you will be fine.
The lateral parts I have seen are the plastic TO-3 case (old guy here, TO-204 these days). The rear side of the heat sinks is not flat. You must mill / file / sand them to provide a suitable mounting surface if you intend to mount a different case style on the rear side.
Personally, I wouldn't even dream of using the laterals. However, this is DIY. Just make sure you understand that by using totally different parts, you are now a design engineer. That is perfectly fine if you know what you're doing.
Hi Kevin,
Yes, measure the resistance between the drains of the two polarities. Also measure from the drain to the source and gate ... after the gate has resistance to the source. Mosfets will store a charge on the gate and change the source to drain resistance between open and very low resistance. ie they can look bad but be perfectly fine. Talking about replacements is premature until you know if they are damaged or not. They could be, but the resistors I see in that case are normally burned a lot worse. So while it's safe and easy to say they are toast right now, you don't get a chance to learn anything at all. Prove they are baked before just accepting it. I test them every single time just to make sure. I've seen enough exceptions to the rule in the last 35 + years, and checking doesn't cost anything more than a little time.
Hi Alan,
The point, "Almost any meter will suffice for servicing these amps. This, like most audio amps, is not a piece of instrumentation." is not reasonable in this age of truly terrible cheap meters I have seen out there. Any meter in tolerance that can be read without too much error is a statement I can agree with.
I think Alan's point is that the bias currents and DC offsets tend to drift around in these amplifiers. This is true, and it takes time and patience to get it adjusted so it doesn't settle way out of whack. A meter that reads way out will ensure that the amp either runs hot and dies - or cold while sounding totally awful. But, if you are going to take the time and trouble to do anything, why use a meter that is telling you lies? Life is hard enough already folks. Occasionally check your meter against a known good one (sort of kind of a cal - just a cross check really). If your meter is out, fix it or replace it and immediately check the new one.
As John mentioned quite correctly, it's that the pairs are matched, not the actual reading showing that match that is important. The lower Vgs number is arguably easier to drive, so lower drive signal and possibly lower distortion. This may also mean an output stage with worse bias stability. That's something that someone will have to check to get an answer.
-Chris
IRF parts no longer available? That's what was used for the SA-100 and SA-220 imported up here.
One word on device matching. The original parts were not matched that well to begin with. I suspect they were just tight enough to do the job. At any rate, use the IRF parts and you will be fine.
The lateral parts I have seen are the plastic TO-3 case (old guy here, TO-204 these days). The rear side of the heat sinks is not flat. You must mill / file / sand them to provide a suitable mounting surface if you intend to mount a different case style on the rear side.
Personally, I wouldn't even dream of using the laterals. However, this is DIY. Just make sure you understand that by using totally different parts, you are now a design engineer. That is perfectly fine if you know what you're doing.
Hi Kevin,
Yes, measure the resistance between the drains of the two polarities. Also measure from the drain to the source and gate ... after the gate has resistance to the source. Mosfets will store a charge on the gate and change the source to drain resistance between open and very low resistance. ie they can look bad but be perfectly fine. Talking about replacements is premature until you know if they are damaged or not. They could be, but the resistors I see in that case are normally burned a lot worse. So while it's safe and easy to say they are toast right now, you don't get a chance to learn anything at all. Prove they are baked before just accepting it. I test them every single time just to make sure. I've seen enough exceptions to the rule in the last 35 + years, and checking doesn't cost anything more than a little time.
Hi Alan,
As you well know, I disagree with you very, very rarely. This will be one of those times we disagree.
The point, "Almost any meter will suffice for servicing these amps. This, like most audio amps, is not a piece of instrumentation." is not reasonable in this age of truly terrible cheap meters I have seen out there. Any meter in tolerance that can be read without too much error is a statement I can agree with.
I think Alan's point is that the bias currents and DC offsets tend to drift around in these amplifiers. This is true, and it takes time and patience to get it adjusted so it doesn't settle way out of whack. A meter that reads way out will ensure that the amp either runs hot and dies - or cold while sounding totally awful. But, if you are going to take the time and trouble to do anything, why use a meter that is telling you lies? Life is hard enough already folks. Occasionally check your meter against a known good one (sort of kind of a cal - just a cross check really). If your meter is out, fix it or replace it and immediately check the new one.
As John mentioned quite correctly, it's that the pairs are matched, not the actual reading showing that match that is important. The lower Vgs number is arguably easier to drive, so lower drive signal and possibly lower distortion. This may also mean an output stage with worse bias stability. That's something that someone will have to check to get an answer.
-Chris
Thanks for the feedback guys.
I am on board for learning as much as I can in this process Chris. I just figured I should pause and share the information about what I found with the gate resistors in case that changed the outlook of things.
My next step will be to put some new resistors back in place and get those other tests done you asked for. Alan said he would try and check for the proper spec for me. Maybe I'll get lucky and won't have to change them again if the Mosfets are still good.
As far as the Toshiba Mosfets are concerned, hopefully I won't need them even if mine are bad. I'm always looking for a back up plan though...can't help it...that whole Murphy's Law bit. I haven't been having much good luck with my audio equipment over the last couple years.
Kevin
I am on board for learning as much as I can in this process Chris. I just figured I should pause and share the information about what I found with the gate resistors in case that changed the outlook of things.
My next step will be to put some new resistors back in place and get those other tests done you asked for. Alan said he would try and check for the proper spec for me. Maybe I'll get lucky and won't have to change them again if the Mosfets are still good.
As far as the Toshiba Mosfets are concerned, hopefully I won't need them even if mine are bad. I'm always looking for a back up plan though...can't help it...that whole Murphy's Law bit. I haven't been having much good luck with my audio equipment over the last couple years.
Kevin
As I stated above, the rear gate resistors (for N channel MOSFETs) were 422 Ohms and the front gate resistors (for P channel MOSFETs) were 165 Ohms. All resistors were 0.5W Roederstein MK3. So any good quality 0.5W resistor would be reasonable. These are E48 decade values.
If you understand the engineering, there is no need for anything like this accuracy - so don't get tied in knots looking for the exact values. I stock a full E24 decade set and therefore use 160R and 430R. These are readily sourced from the likes of Digikey, Mouser etc
If you understand the engineering, there is no need for anything like this accuracy - so don't get tied in knots looking for the exact values. I stock a full E24 decade set and therefore use 160R and 430R. These are readily sourced from the likes of Digikey, Mouser etc
I'll try again...
A soundcard is good enough to do signal tracing. As in - if I poke a 1kHz sine wave into the amp how far does it get? It will tell you if the amp is working but not very much about how well it is working because the bandwidth is so limited. And you need to protect the inputs to prevent damage to the soundcard - the amp output can have an 80V swing and the soundcard is toast at much above 2V.
A good soundcard is useful for doing THD measurements, so if you are keen to get the most out of your soundcard for this purpose, do a search for Pete Millets (sp?) soundcard interface project here on diyaudio.
At any rate...after some quality time this afternoon with the amp and soldering iron, I successfully placed some new resistors at the N-type gates of the left channel where the originals were toasted. I used what I had on hand, the closest I could get to spec was a .5 watt 390 Ohm resistor. Getting the lead from the main board for the gates back in place was a major PITA.
I'm gonna just list the testing I did and give explanations along the way. I rechecked myself about 6 times due to things I discovered along the way and my OCD. My assumption when I started these tests was that the polarity of the meter should be applied to mimic what I thought was appropriate to what I was seeing with the Mosfets. I assumed that because the "Drain" was anchored to the heat sink that it should be probed with the negative or black probe from my meter. I didn't think polarity was an issue between the drains. Later on I discovered that polarity did seem to affect some of my tests, which led to more re-testing. I did a few probes before deciding that the scale to use on my meter that seemed most appropriate maxed out at 2000 Ohms. I noticed that it seemed like I was charging something in the circuit while performing these tests, so I made it a habit to put my resistor rig across the power caps to discharge them in between tests to be safe and hopefully more accurate.
Resistance between N-type and P-type Drains
Channel that had damaged N-type gate resistors (left): momentary negative reading, climbs and tapers off slowly to a maximum resistance of about 670 Ohms.
Channel with no apparent damage (right): momentary negative reading, climbs steadily until the resistance passes the 2000 Ohm limit the meter is set to.
Resistance between Drain and Gate
All these measurements seemed to register negative momentarily on the meter when first probed as well.
Left Channel
N-type (had damaged resistors)
Upper- jumps to about 70 Ohms, then slowly creeps to level off at 77 Ohms
Lower- jumps to about 50 Ohms, then slowly creeps to level off at 55 Ohms
P-type (no apparent damage)
Upper- jumps to about 700 Ohms and then steadily climbs until it exceeds the 2000 Ohm setting on the meter
Lower- same behavior as the upper P-type in this channel
Right Channel
N-type (no apparent damage)
Upper- jumps and then steadily climbs to level out at 625 Ohms
Lower- same behavior as upper
P-type (no apparent damage)
Upper- jumps to about 700 Ohms, then steadily climbs until it exceeds the 2000 Ohm setting on the meter
Lower-same behavior as the upper
I rechecked all the Drain to Gate measurements after making the Drain to Source measurements to see if the polarity of the probes changed any readings. The only thing I noticed was that the measurements for the P-type Mosfets started off with a negative value and then climbed toward zero and then continued to climb past the positive 2000 Ohm setting. I believe the N-type Mosfets would not read with the probes reversed (read, I was getting tired).
Resistance between Drain and Source
Somewhere during the process of taking these measurements I got confused by inconsistency between what I thought should be two functional sets of P-type Mosfets in the Left and Right channels with measurements that should approximately mirror one another. I must have scratched my head and re-taken these measurements at least 4 times before realizing that polarity had an affect on what I was measuring between the Drain and Source, I must have reversed the probes once by accident. I have listed the associated measurement and polarity that appeared to reflect usable data here. When attempting to reverse polarity from that which is noted for these figures I consistently ended up with a resistance measurement that climbed off the scale of the 2000 Ohm setting on my meter.
Left channel
N-type (had damaged resistors)
Upper- Red probe to source: No resistance registered
Lower-Same behavior as upper
P-type (no apparent damage)
Upper- Red probe to drain: 434 Ohms
Lower- Same behavior as upper
Right channel
N-type (no apparent damage)
Upper- Red probe to source: 445 Ohms
Lower- Same behavior as upper
P-type (no apparent damage)
Upper- Red probe to drain: 438 Ohms
Lower- Same behavior as upper
Well that's it. It's looking pretty much to me like there is something definitely wrong with those N-type Mosfets in the left channel that had fried resistors. I actually tried to read up a bit on how Mosfets work so I could understand more what I was testing...I don't think I found the appropriate material...semi-conducting surfaces and etc. are just over my head, or just of no interest to me on that level. I'll have to be happy I have you guys to help for now.
Kevin
Hi Kevin,
With any transistor, the polarity of the applied test voltage can really change the readings. As long as you have some resistance between the source and gate, you shouldn't have much to worry about from a stored charge on the gate.
If you're careful, you can use stored gate charge to test a mosfet transistor. That's something you should definitely read a little on or you might get a bit confused by the readings. Anyway, you can charge the gate one way with the ohmmeter, then read a low resistance between the drain and source. Charge the gate the other way and read an open between drain and source. Pretty cool stuff.
Red can be hard on the eyes.
-Chris
With any transistor, the polarity of the applied test voltage can really change the readings. As long as you have some resistance between the source and gate, you shouldn't have much to worry about from a stored charge on the gate.
If you're careful, you can use stored gate charge to test a mosfet transistor. That's something you should definitely read a little on or you might get a bit confused by the readings. Anyway, you can charge the gate one way with the ohmmeter, then read a low resistance between the drain and source. Charge the gate the other way and read an open between drain and source. Pretty cool stuff.
Red can be hard on the eyes.
-Chris
Kevin,
You should note that the Toshiba N- and P-type are TO-3P (plastic) case vs. the TO-3 metal case originals. You have to plan really well how to properly mount/wire these and each adjacent part before deciding to go with the Toshiba mosfets.
You must also troubleshoot what's causing the original mosfets to fail before putting in the fresh ones...or they'll just burn up again.
You should note that the Toshiba N- and P-type are TO-3P (plastic) case vs. the TO-3 metal case originals. You have to plan really well how to properly mount/wire these and each adjacent part before deciding to go with the Toshiba mosfets.
You must also troubleshoot what's causing the original mosfets to fail before putting in the fresh ones...or they'll just burn up again.
Here is a shematic for the SA100 ... i don't know where it came from and it's in spanish.
HTGuide Forum - View Single Post - Counterpoint Amp Repair
Well, here is an output stage that I have drawn up. No sims or anything. I used the BJT parts that Chris recommended. Any thoughts on this, before I try a simulatution?
The driver stage is biased into class A up to 100watts RMS in to 8 ohms at 100mA. And I just kind of guess with the base stoppers. I suspect this would take some playing around with to get it just right.
I would probably take Chris's advice and add 5-6dB of feedback as this amp has about 30dB of gain.
So, there it is.
Chris, the IRF130 and IRF9130 parts are available at a price. Newark has the IRF9130 for $9.00 each and another place has a sale on the IRF130 and can be had for $42.00 ... that's $10.00 off.
But, we can always go with other MOSFETs.
HTGuide Forum - View Single Post - Counterpoint Amp Repair
Well, here is an output stage that I have drawn up. No sims or anything. I used the BJT parts that Chris recommended. Any thoughts on this, before I try a simulatution?
The driver stage is biased into class A up to 100watts RMS in to 8 ohms at 100mA. And I just kind of guess with the base stoppers. I suspect this would take some playing around with to get it just right.
I would probably take Chris's advice and add 5-6dB of feedback as this amp has about 30dB of gain.
So, there it is.
Chris, the IRF130 and IRF9130 parts are available at a price. Newark has the IRF9130 for $9.00 each and another place has a sale on the IRF130 and can be had for $42.00 ... that's $10.00 off.
But, we can always go with other MOSFETs.
Attachments
Hi John,
You can't implement that output stage as you've drawn it. The On Semi parts are BJT types firstly, so nowhere near a 1:1 swap. You will need a bias control circuit (there is one already, but the thermal gain may be off). You also need a buffer between the voltage amp and the outputs. Tube outputs are not going to like a highly variable input impedance like one presented by a standard triple either.
I used a power diamond buffer. When you've got it working right, the best sound quality I've heard. When it isn't working right, it sounds "off". Also, it will current limit depending on how you set it up. The core of my output stage is a diamond buffer, but there is more and I'm not done yet. The best part is that my worst case DC offset is +/- 35 mV - without adjusting anything!
As for the original parts from IRF, they have gone up a lot. So ... use a more current substitute from them. Those parts are admittedly long in the tooth by now. So, just find a modern substitute.
As for the plastic TO-204 devices, I mentioned recently (A couple posts ago) what you need top what for, and where I know there is a problem. Remember, I set up an alternative BJT output stage on the same heat sinks. I used the same case style as the originals. You have to love On Semi for making the same number available in more than one package.
-Chris
You can't implement that output stage as you've drawn it. The On Semi parts are BJT types firstly, so nowhere near a 1:1 swap. You will need a bias control circuit (there is one already, but the thermal gain may be off). You also need a buffer between the voltage amp and the outputs. Tube outputs are not going to like a highly variable input impedance like one presented by a standard triple either.
I used a power diamond buffer. When you've got it working right, the best sound quality I've heard. When it isn't working right, it sounds "off". Also, it will current limit depending on how you set it up. The core of my output stage is a diamond buffer, but there is more and I'm not done yet. The best part is that my worst case DC offset is +/- 35 mV - without adjusting anything!
As for the original parts from IRF, they have gone up a lot. So ... use a more current substitute from them. Those parts are admittedly long in the tooth by now. So, just find a modern substitute.
As for the plastic TO-204 devices, I mentioned recently (A couple posts ago) what you need top what for, and where I know there is a problem. Remember, I set up an alternative BJT output stage on the same heat sinks. I used the same case style as the originals. You have to love On Semi for making the same number available in more than one package.
-Chris