That's why I never complained , I knew of this.
Oh , the joys of this modern age.
OS
We used to print the schematic on the inside of the top cover. We decided to stop because we would have to scrap up to 50 lids (the normal batch size then, now 100) if we made a change to the circuit. We figured that anybody could just trace it out if they really wanted to, plus if someone needed service in twenty years and we had gone out of business (we were just starting up), that a good tech could fix it from the schematics.
It turned out to be not such a good idea. There isn't enough information in the schematic to properly set the DC offset or the bias. Even we didn't work out all the bugs until about six years after we had discontinued the model! Every once in a while we would get a unit back, repair it, and it would fail very quickly again. So we knew something was going on but we didn't know what.
We sat down and went through everything with a fine-toothed comb. Eventually we found that the IR datasheets are garbage. When you put the part in a curve tracer, the way it actually behaves is very different from what the datasheet says. Even though it is a 60 volt part, when Vds goes above 30 volts, Id increases markedly.
The nominal rail was 25 volts, but if the AC line was high then the rail would hit 30 or 32 volts. Then the idle current would go up and it would go into thermal runaway and self destruct. It took forever to find a part that behaved correctly but now they are totally reliable.
Really ? We have discussed this circuit in the "Need to build JC 2 preamp",
and I would say this ebay version has to low current on the input fets and
even worse on the two bipolars. (This is probably why they added a buffer).
I´ve built the circuit like we discussed it in the thread mentioned and it measures and sounds excellent. Not so sure about this clone on ebay.
Charles, Ed Simon and I have discussed the circuit, off line. We all agree that this is one heck of a deal and can get you 75% there just with what they 'promise'. Then, this circuit could be modified to be many things, including a low power amp, headphone amp, or to run open loop. WOW! Such a deal!
Well OT but it will get there. A few weeks back Bill M. the AV faculty from a college south of here called, he had purchased a Hammond organ/Leslie and wanted help moving them to a local guy who was the high guru of Hammonds. I turned him over to Brian who we let use about 1000 sq ft of space here for his backline business (Rental of band gear to touring bands... hard to get a Hammond on a plane.) Brian helped him out and met the guru and was so impressed he asked him to come and go through some of his gear. The guru not only had been an electronic tech for 40+ years he also played keyboards in several bands over the years.
So yesterday he was scheduled to come in. Just before he arrived all of us here noticed a burning smell in the back shop. I turned of the HVAC units and checked them both for overheating motors, no luck. We went around unplugging everything we could find, no luck. I had Warren remove all the cardboard boxes we save for reuse. Still nothing. The guru arrived and helped look for the source. The odor kept getting stronger, but no smoke! Finally I decided to try putting water in the floor drain, as it really never gets used, the trap could dry out. That seemed to do it. So it seemed there must have been a fire and much of the debris made it into the drain. Two weeks ago the gas company fixed their lines that had been leaking since 2004. After that John called and he mentioned he used an IR thermometer to look for hot spots. So the first lessons are no matter where you look and how logically there can always be a surprise answer. Second John actually has good ideas! Third I really don't like the local gas company!
Part two. The guru was surprised by my shop in particular the tube amp set up on my test bench that was being used for the distortion measurements shown here. He started to go over the organs. The Leslie had a motor installed upside down.. The Wurlitzer need a reed trimmed etc. At one point to trim the reeds he needed a reference tone. I suggested using an oscillator and frequency counter. He tried it but the resistance tuned oscillators I had drifted too much. So he used another electronic keyboard.
Afterwards we discussed bits and pieces and one of the things I do is show folks the office stereo. It uses corner horns as shown in AX. I always note how far away folks stand from the speakers while listening to them. Studio guys get too close to hear the stereo image, I have set the sweet spot to be 27 feet from them. The guru admitted he did not really understand what was meant by stereo imaging! A nice admission from someone who pretty much only listens to live music. I tried to show him but he really did not hear the difference. All the other details he seemed to hear at a level very few do.
So the next issue is how many actually know where their loudspeakers sweet spot is and where it is aimed? How many have checked to see if their loudspeakers are even in a good spot? So even among experienced listeners there will be differences in experience and preferences.
Final lesson is what would have happened if the gas company lines were still leaking?
So yesterday he was scheduled to come in. Just before he arrived all of us here noticed a burning smell in the back shop. I turned of the HVAC units and checked them both for overheating motors, no luck. We went around unplugging everything we could find, no luck. I had Warren remove all the cardboard boxes we save for reuse. Still nothing. The guru arrived and helped look for the source. The odor kept getting stronger, but no smoke! Finally I decided to try putting water in the floor drain, as it really never gets used, the trap could dry out. That seemed to do it. So it seemed there must have been a fire and much of the debris made it into the drain. Two weeks ago the gas company fixed their lines that had been leaking since 2004. After that John called and he mentioned he used an IR thermometer to look for hot spots. So the first lessons are no matter where you look and how logically there can always be a surprise answer. Second John actually has good ideas! Third I really don't like the local gas company!
Part two. The guru was surprised by my shop in particular the tube amp set up on my test bench that was being used for the distortion measurements shown here. He started to go over the organs. The Leslie had a motor installed upside down.. The Wurlitzer need a reed trimmed etc. At one point to trim the reeds he needed a reference tone. I suggested using an oscillator and frequency counter. He tried it but the resistance tuned oscillators I had drifted too much. So he used another electronic keyboard.
Afterwards we discussed bits and pieces and one of the things I do is show folks the office stereo. It uses corner horns as shown in AX. I always note how far away folks stand from the speakers while listening to them. Studio guys get too close to hear the stereo image, I have set the sweet spot to be 27 feet from them. The guru admitted he did not really understand what was meant by stereo imaging! A nice admission from someone who pretty much only listens to live music. I tried to show him but he really did not hear the difference. All the other details he seemed to hear at a level very few do.
So the next issue is how many actually know where their loudspeakers sweet spot is and where it is aimed? How many have checked to see if their loudspeakers are even in a good spot? So even among experienced listeners there will be differences in experience and preferences.
Final lesson is what would have happened if the gas company lines were still leaking?
Last edited:
Thanks again for sharing that info. I do not use much of IRs, only IRF610/9610 as cascodes, at 60V, and about 10mA. As VAS, I use Hitachi 2SK413/2SJ118, and they never failed. And output stage is bipolar, with 8 pcs MJL1302/3281.
Every single P-channel MOSFET IR has ever made has a serious defect. The transconductance changes by a factor of ~2 after about a millisecond. This puts it right in the middle of the audio band (1 kHz). If you run the devices very hot (above 60 C or so case temperature, hot enough to burn your hand), this effect subsides.
Another thing that most people miss is that the IRF9xx0 is not the correct complement to the IRFxx0. It may be in terms of switching abilities (which is what they are all designed for, but not at all for linear amplifiers. The N-channel part will have significantly higher transconductance than the P-channel part.
The easiest rule of thumb is to use a 200 volt N-channel part and a 100 volt P-channel part. (But PLEASE use another brand besides IR for the P-channel parts!) You are using much higher voltage parts, so you will have to put some different parts on the curve tracer. Maybe a 400 volt P-channel part with the same current rating as an 800 volt N-channel part will give similar transconductance.
But no matter what you do, you will still be stuck with the non-linear input capacitance of the vertical MOSFET. Since you are not using them as output devices, then I think you will have much better sonic results by replacing the vertical MOSFETs with lateral parts from Exicon, Magnatec, Class-D, or ALFet (all made by Semelab).
Most of the mosfets were designed for switching applications so you do not get ideal linear behavior. The gate capacitances (Cgs and Cgd) are very non linear wrt to both Vds and applied gate voltage. So, MOSFET selection has to be done carefully, and drver circuits do matter even in linear applications. Maybe Bob Cordell could comment further.
Last edited:
Of course it is helpful to hear it yet again , but this is why the amp builders go N-channel quasi http://www.diyaudio.com/forums/solid-state/43331-power-amp-under-development.html for V-mosfet OP stages. So even the "amateurs" know about the P/N channel shortfall's , some even consider designs like this as barely "mid-fi", if not outright dangerous. The internet is also chock full of dangerous V-mosfet designs to attract the "raw power" crowd , enticing them to hook up the highest amperage- 80V rail PS's to many a ill conceived circuit. In short - "newbies" playing with 160V + 100kuf in order to get the party going.
The slightly advanced amateur has already accepted the lateral mosfet as the "only way". They even have different P/N Cgs and need to be (gate)stoppered accordingly.
PS - I DO hope more will read Cordell's chapters on mosfets = better, safer designs.
Edit - in no way did I imply that the above linked project is a "bomb" , but great care has to be taken with the layout and power/voltages involved. Some (projects) have oscillated and "expired" (VHF oscillation).
OS
Last edited:
In general, complementary output devices, including fets, or bipolars are close enough in their actual analog transfer, both in time and linearity to have an advantage over quasi-complementary, or other design approaches that eliminate complementary output devices. Charles found this problem with the IR fets, and virtually all hi end designers know about it now, but it IS conveniently lost in high feedback amplifiers.
This is one of the things that expert designers have to contend with: We have to use the devices AVAILABLE to us, and somehow make them work as good as possible in our circuits.
This is one of the things that expert designers have to contend with: We have to use the devices AVAILABLE to us, and somehow make them work as good as possible in our circuits.
Old information. The transconductance of the P ch parts declines several
dB in the 500 - 1KHz region and shelves flat above that. It simply appears
to shift after a millisecond if you are looking at it on a curve tracer. This
is documented in my article on Mosfet testing which is posted online, and
has been discussed at length in these forums.
By my opinion much better assortment than IR have IXYS IXYS Product Portfolio | Power Devices ...you can find there types, which have closer complementary parameters...
- Status
- Not open for further replies.