Hi,
I don't often plead for help, but I've got a Counterpoint SA-100 with a blown channel that's been sitting around for a few years. I finally decided to get it up and running again and purchased a full compliment of Exicon 20N16 and 20P16 to replace the CPT10N12 & 10P12.
I don't know why, but I seem to have damaged the Exicons between the gate and sources - both the N & P types.
I did this only on one channel because the other channel had a working set of the original MOSFETs. Just for a sanity check, I insalled a working set of devices from a Counterpoint SA-220 (RFM 10N15 and IRFP 9241) in the blown channel and it works fine - so I know for sure both channels work and that the issue is with the MOSFETS.
I thought that by slowly bringing the amp up on a variac wold keep me from damaging the Exicons in case there were some bias voltage problems. I brought the rails only up to +/- 10V or so. The gate voltage went up to around 1.5V on the N side, and suddenly went way down. I noticed this morning that the 1A safety fuse I installed in the rail was blown, so I replaced it. The other rail fuse had a 1ohm resistor so I could monitor current. I just cracked the variac open (for 300mv at rails) after replacing the fuse and now the devices are drawing as much current as they can - this means a short somewhere. I measured the N and P MOSFETs with the diode tester and I'm getting strange numbers.
What did I do wrong? Am I mis-handling the Mosfets somehow? I'm totally not MOSFET savvy so it may be something really rudimentary that I neglected to do. I have one more channels worth of Exicons, but I don't want to destroy these too.
I know Chris might say that I'm abusing good MOSFETS by installing them in a Counterpoint, but I really just want this to work without having to re-engineer it, and the Exicons were (I thought) the easiest way to get it to work.
Thanks for your help,
I don't often plead for help, but I've got a Counterpoint SA-100 with a blown channel that's been sitting around for a few years. I finally decided to get it up and running again and purchased a full compliment of Exicon 20N16 and 20P16 to replace the CPT10N12 & 10P12.
I don't know why, but I seem to have damaged the Exicons between the gate and sources - both the N & P types.
I did this only on one channel because the other channel had a working set of the original MOSFETs. Just for a sanity check, I insalled a working set of devices from a Counterpoint SA-220 (RFM 10N15 and IRFP 9241) in the blown channel and it works fine - so I know for sure both channels work and that the issue is with the MOSFETS.
I thought that by slowly bringing the amp up on a variac wold keep me from damaging the Exicons in case there were some bias voltage problems. I brought the rails only up to +/- 10V or so. The gate voltage went up to around 1.5V on the N side, and suddenly went way down. I noticed this morning that the 1A safety fuse I installed in the rail was blown, so I replaced it. The other rail fuse had a 1ohm resistor so I could monitor current. I just cracked the variac open (for 300mv at rails) after replacing the fuse and now the devices are drawing as much current as they can - this means a short somewhere. I measured the N and P MOSFETs with the diode tester and I'm getting strange numbers.
What did I do wrong? Am I mis-handling the Mosfets somehow? I'm totally not MOSFET savvy so it may be something really rudimentary that I neglected to do. I have one more channels worth of Exicons, but I don't want to destroy these too.
I know Chris might say that I'm abusing good MOSFETS by installing them in a Counterpoint, but I really just want this to work without having to re-engineer it, and the Exicons were (I thought) the easiest way to get it to work.
Thanks for your help,
Yeah, but with 10V source to drain and 1.5V at the gate I would imagine the MOSFET would be perfectly happy with this.ilimzn said:
Is that the difference - I really don't know.Leolabs said:
According to the Altavista Audio site, the exicon MOSFETs can be used. I understand that they need to be biased differently than the original MOSFETs, but I thought I was being careful by limiting the voltage.
Someday I'll get this working...
woodman said:
How did you handle your MOSFETs? You have to remember that the insulation between the gate and substrate is very thin, has a whoppingly high impedance, and, therefore, is highly susceptable to damage from static electricity. Especially during wintertime when dry air is dried even more by central heating. It doesn't take very much, after all, to see and feel a spark means that you have several thousand volts. It takes way less to ruin a MOSFET. Furthermore, these don't very often poof at once, but rather have partial gate damage that only gets worse with time.
Tp prevent that, it's essential to not remove the shorting collars from the leads until ready to install, make sure you use a grounding strap when you handle them, make certain that the equipment is grounded and that there is no residual charge on PS filter capacitors.
Did you do that?
Hi James,
Other things to be aware of:
1) The relay shorts the gates to ground until the circuit times out. You can pull the relays for testing.
2) You may need to change the gate snubber resistor values.
3) Check the darn zeners. Do you need a different value?
4) Most important. Do not tighten the devices until the sockets almost break as recommended in the service manual.
You may be able to get away with a lower bias current than teh original parts. This way you can still use the bias circuit (with adjusted values for lower gate voltage). You may need to mess with the bias emitter resistor to adjust the correction.
Now that you've done this much, do the other channel.
Fix the voltage amp stage design issues. You will have to match these outputs very closely as you do with the IR parts.
-Chris (am I really that infamous?? )
Well, yes. I imagine these would sound better than the IR devices once they are connected properly and biased properly.
Other things to be aware of:
1) The relay shorts the gates to ground until the circuit times out. You can pull the relays for testing.
2) You may need to change the gate snubber resistor values.
3) Check the darn zeners. Do you need a different value?
4) Most important. Do not tighten the devices until the sockets almost break as recommended in the service manual.
You may be able to get away with a lower bias current than teh original parts. This way you can still use the bias circuit (with adjusted values for lower gate voltage). You may need to mess with the bias emitter resistor to adjust the correction.
Now that you've done this much, do the other channel.
Fix the voltage amp stage design issues. You will have to match these outputs very closely as you do with the IR parts.-Chris (am I really that infamous??
)Re: Re: Counterpoint SA-100 with Exicon MOSFETs
Umm...no. I stupidly thought that the protection diode exempted me from practicing any of the ESD training that I've had for the past 10years. I'll have to bust out the wrist strap now. Also, I bought the Exicon's on Ebay and they were unceremoniusly placed in ESD bags without the black conductive foam - they could have been bad before I got to them....
Miles Prower said:
Umm...no.
I stupidly thought that the protection diode exempted me from practicing any of the ESD training that I've had for the past 10years. I'll have to bust out the wrist strap now. Also, I bought the Exicon's on Ebay and they were unceremoniusly placed in ESD bags without the black conductive foam - they could have been bad before I got to them....anatech said:Hi James,
Well, yes. I imagine these would sound better than the IR devices once they are connected properly and biased properly.
Other things to be aware of:
Yup, did that.
Did that too - though I used the Moscode 300 output stage as a guide.
Zeners were good. They're 10V zeners and I probably needed a smaller value.
I only tighten them as much as they need to be.
I probably should have done this but I figured limiting the power via the variac would protect the MOSFETs.
Match them? I thought lateral MOSFETs didn't need to be matched. Besides I was gonna hang some ballast resistors on the sources - would that have worked?Now that you've done this much, do the other channel.
-Chris (am I really that infamous??
Famous, not infamous
BTW - let me know when you're willing to share your BJT SA-100 mod.
Hi James,
-Chris
I'm sure that you are aware how tight transistors ought to be. Most of what I posted was for the benefit of others. Wasn't sure you knew about the relays.
Should be close enough for Rock & Roll!
Actually, no. If you increase the AC voltage slowly you may reach a point where they oscillate or conduct heavily. They then discharge all the energy in the filter caps. Insta poof! An open gate resistor will do this.
I'm not sure on that. I normally err on the side of caution. I am use to matching transistors on everything. We had to match our own mosfets for Adcom service. I'm betting many didn't do that because I asked for the procedure and it took a while to get it.
Not with this design. Ballast resistors will really adversely affect the sound quality. There is no feedback from the output.
I've made huge gains in the sound quality but I'm still not happy. Work has been slow due to a head injury about 1 1/2 years ago. I'm still trying to learn what doesn't make sense anymore. I do have prototypes made and I'm playing with them. Changes to every circuit, but you can still use the PCB. I use bipolars for outputs instead. They still have the Counterpoint sound, but much cleaner, cooler and more stable. Forgot quieter and better sound. SY heard one.
-Chris
djk said:
I'll re-quote this here. S and D are exchanged on lateral with respect to vertical MOS, did you check this?. Laterals also have the D-S protection diode which would present a near short if S anad D are reversed. Also, and unfortunately, I know this from experience, forgetting to reverse S and D in a vertical to lateral MOS conversion, will usually destroy the lateral FETs, the reverse protection diode can fry.
...one year later...
Well, I finally got this amp running with the Exicons!
I decided to redouble my efforts in getting this amp working over the holiday season. The first thing I did was to re-check the MOSFETS that I tried using last year and they were not damaged at all! I didn't understand that the gates could be charged with the voltmeter, so I was getting strange measurements (because they're not BJTs).
After installation, I shorted the gates to ground and applied power with no problems, so I knew the devices were OK.
The bias voltage in the counterpoint was way too high for the Exicons, so I changed the bias circuit from a vbe multiplier to a non-temperature compensated network. I adjusted until I saw a 250ma current draw at the rail. It was so nice to finally see the results I expected!
Without double checking anything and just to satisfy my curiousity, I appled signal and got a nice clean looking 100W sine! Woo Hoo!
Now I'm going to go back to dot my 'I's and cross my 'T's. It works, but now needs some final touches.
I would like to thank everyone who responded to this post - your help was essential and much appreciated!!
Well, I finally got this amp running with the Exicons!
I decided to redouble my efforts in getting this amp working over the holiday season. The first thing I did was to re-check the MOSFETS that I tried using last year and they were not damaged at all! I didn't understand that the gates could be charged with the voltmeter, so I was getting strange measurements (because they're not BJTs).
After installation, I shorted the gates to ground and applied power with no problems, so I knew the devices were OK.
The bias voltage in the counterpoint was way too high for the Exicons, so I changed the bias circuit from a vbe multiplier to a non-temperature compensated network. I adjusted until I saw a 250ma current draw at the rail. It was so nice to finally see the results I expected!
Without double checking anything and just to satisfy my curiousity, I appled signal and got a nice clean looking 100W sine! Woo Hoo!
Now I'm going to go back to dot my 'I's and cross my 'T's. It works, but now needs some final touches.
I would like to thank everyone who responded to this post - your help was essential and much appreciated!!
Yer making me want to 'bust a move' and break out one of my counterpoints and fix it. I accidentally blew it up after Chris fixed it! Damn!
And Chris, if you are still interested in going forward with your 'byrd design, I still have that spare SA-220 chassis that you can use. I'm never going to do anything with it. I'll pull the working outputs and ship it over.
And Chris, if you are still interested in going forward with your 'byrd design, I still have that spare SA-220 chassis that you can use. I'm never going to do anything with it. I'll pull the working outputs and ship it over.
Hi Ken,
I've been knocked on my tail for a little over a year now. "They" are changing one of my meds right now, so I hope to become more useful in the near future.
Take care Ken, Chris
How the devil did you do that? Oh well. We can have another go at it if you want. That one isn't the one I matched the FETs for, would it? Or, were these the bad FETs that Mike shipped up and we had to wait for a good set? I can't remember the details. If it's the one I picked and matched the FETs for, I'm gonna cry! Those were hand matched to much tighter tolerances that anything out of his factory. I broke up some matched sets of Adcom outputs to obtain that crazy tightness in specs.
Yes I am. Slowly and unshurely (i doubt that's a word!). I also have some more ideas for tube gain stages that I'll also be trying out. The beauty of the design is how modular it is. You can swap gain stages for different types - that's the idea anyway.
Really?! Cool! Thank you very much for that. You are also welcome to swing by whenever it's convenient for you. It would be good to see you again some time, if you have the time.
I've been knocked on my tail for a little over a year now. "They" are changing one of my meds right now, so I hope to become more useful in the near future.
Take care Ken, Chris
Hi woodman,
How hot are your heat sinks running now? Do you have any way to measure your distortion levels now compared to what they were? I have a feeling the Exicons may sound better.
-Chris
That is expected, given the differences between the devices.
I'm not so sure I am comfortable with that. You need to check how evenly the mosfets are sharing current in that amp as well. No source resistance at all, so it's all up to your matching.
How hot are your heat sinks running now? Do you have any way to measure your distortion levels now compared to what they were? I have a feeling the Exicons may sound better.
-Chris
anatech said:
Hey Chris - it's good to hear from you! Are you still recovering from your head injury?
OK, so I played with this amp all weekend and here's what I've got:
-Bias seems to be dead-nuts stable no matter how hot the heatsinks run. Temp is around 100deg F at idle and it stays that way for hours.
-THD is generally 0.2-0.3% lower than the original specs (and without weighting) - I don't like the way the distortion output looks, but it's probably because the output stage has no feedback.
-I have not yet measured for current hogging, but I keep reading conflicting info regarding lateral MOSFETS. Some say they don't require matching because of the tempco, others say they do. I have not seen any strange behavior or hotspots on the heatsinks, so I'll wait to look at this more closely.
-I'm not quite getting 100W after all, more like 90W.
-I'm only applying around 500mv to the gates to set the output current to 250mA.
I was feeling pretty good about the amp last night, so I removed the 1 ohm resistor from the +rails (for setting bias current) and inserted the fuse in preparation to listen through my test speakers.
With the fuse installed, the output does not look good at all - there is an almost square-wave type bump in the output. I didn't measure the frequency of the bump, but it looks like it's line/rectifier related with maybe a 30% duty cycle. If I put the resistor back in the rail (+ or -), everything looks fine. I have not yet had time to look into why this is happening.
The resistor in the + or - rail does affect the clipping - wherever the resistor is, the corresponding edge of the waveform is affected.
I feel like I'm close though
Cheers,
Hi James,
I'm slightly better, and worse in other ways, thank you for asking.
-Chris
I'm slightly better, and worse in other ways, thank you for asking.
Seems a bit warm, but that is the "happy spot" for these outputs. I'd leave it alone now. Just don't over tighten the outputs as the service manual suggests.
I'm not surprised at all. Better output devices will make a large gain in THD performance. Simply changing to bipolars has the same effect you are seeing. You are on the right track, perfectly valid.
No kidding! Match them. Performance should improve. Stick a 1 ohm resistor in each drain and measure to see what your current sharing is like. This would probably have more effect around the crossover region, for the better and especially with the circuit you are working with.
Interesting. Your output power should really be slightly higher than original. You have more headroom with a lower gate - source voltage. This difference should be slight, but certainly not the other way. Your rails should be measuring 50 VDC. The earlier SA-12 had rails around 45 VDC. What do you have there?
Look at your heater supply voltages with an oscilloscope. The heater capacitors often fail due to high ripple current and poor cooling. Also look at the dropping resistor. While you are at it, look at the ripple on your HV supply. Be careful with your 'scope as that will run around 300 ~ 330 VDC. AC coupling, use an external blocking capacitor if your 'scope is a USB or digital type. If your B+ caps are failing, your amp could begin to motorboat. This design is on the verge of instability and degraded capacitors will push it over the edge.
The two problems are probably related. Also, when doing higher power tests, the one ohm resistors must be removed. I bought some current probes to study supply current problems. If you find an old HP 428B and fix it up, you will be able to investigate this further. The new Agilent current probes are out of my price range, and so are the Tek models. Be aware these old HP units are tube technology, but they work very well. I am pleased with what mine have shown so far. I will therefore restore them, they are well worth the effort.
If you go any further, you will run into the issues with his circuit design and poor power supply. Keep an open mind as you look at these things. There are issues that need to be resolved before you can go any further after you get your current problems solved.I feel like I'm close though
-Chris
Well, it took awhile to understand what I was looking at on the scope. I was getting weirdness that indicated a problem, but did not point to the nature of the problem.
It ends up that my distortion analyzer was outputting strange things to the 'scope due to the very high frequency oscillation (~25MHz !) of the MOSFETS, as measured with a scope probe.
Now that I have something to work with, I'll try different gate resistors, ferrite beads, tighter layout and local bypassing and see if it clears up. I'm also going to go back and re-do the bias circuit either with a thermistor or VBE multiplier as it looks like I will need temp compensation after all.
I haven't looked at MOSFET matching/current sharing yet as there is not much I would be able do with the devices I have if they don't happen to match well enough. I may end up using different devices anyway - I can get 10N16/10P16 matched on ebay.
The good thing is that the 20N16/20P16 can still be used to realize my daydream of building a MOSFET output Adcom 555!
Cheers,
It ends up that my distortion analyzer was outputting strange things to the 'scope due to the very high frequency oscillation (~25MHz !) of the MOSFETS, as measured with a scope probe.
Now that I have something to work with, I'll try different gate resistors, ferrite beads, tighter layout and local bypassing and see if it clears up. I'm also going to go back and re-do the bias circuit either with a thermistor or VBE multiplier as it looks like I will need temp compensation after all.
I haven't looked at MOSFET matching/current sharing yet as there is not much I would be able do with the devices I have if they don't happen to match well enough. I may end up using different devices anyway - I can get 10N16/10P16 matched on ebay.
The good thing is that the 20N16/20P16 can still be used to realize my daydream of building a MOSFET output Adcom 555!
Cheers,