I just rebuilt one channel of a very large Crest CA12 power amp. this amp is 700w per channel at 8 ohms.
It is a class H rail switch design using 6 pairs of 2SC3281/2SA1302 transistors for the amp section and 5 pairs for the rail switch section.
The 2SC3281's are no longer available and Crest AKA Peavey recommends replacing them with MJL21194/3 devices. This is what there own repair shop does.
I rebuilt one channel and it works perfectly fine. However i noticed that the rebuilt channel draws more AC Line current for a given output than the stock channel.
For example. the amp idles around 2amps as measured by an AC current meter on my variac. with an 8 ohm load connected one channel only.
20Vrms into 8 ohms stock channel draws about 3.5 amps.
the rebuilt channel 20vrms into 8ohms draws about 5.5 amps from the ac line.
everything in the amp channel is exactly the same with the exception of the output devices. Both channels are biased exactly the same. No oscillation is observed. So why the additional current draw? I mean it is considerable.
at 50Vrms out on the rebuilt channel it is drawing 10 amps from the AC power line where as the stock channel is around 7.
Both channels clip at 80vrms into 8 ohms one channel driven. Both have the same gain.
What is it about the specs of these two transistors that is different enough to cause this behavior???
It is a class H rail switch design using 6 pairs of 2SC3281/2SA1302 transistors for the amp section and 5 pairs for the rail switch section.
The 2SC3281's are no longer available and Crest AKA Peavey recommends replacing them with MJL21194/3 devices. This is what there own repair shop does.
I rebuilt one channel and it works perfectly fine. However i noticed that the rebuilt channel draws more AC Line current for a given output than the stock channel.
For example. the amp idles around 2amps as measured by an AC current meter on my variac. with an 8 ohm load connected one channel only.
20Vrms into 8 ohms stock channel draws about 3.5 amps.
the rebuilt channel 20vrms into 8ohms draws about 5.5 amps from the ac line.
everything in the amp channel is exactly the same with the exception of the output devices. Both channels are biased exactly the same. No oscillation is observed. So why the additional current draw? I mean it is considerable.
at 50Vrms out on the rebuilt channel it is drawing 10 amps from the AC power line where as the stock channel is around 7.
Both channels clip at 80vrms into 8 ohms one channel driven. Both have the same gain.
What is it about the specs of these two transistors that is different enough to cause this behavior???
If I hear you right, you're talking a 250W difference in mains power draw when running a 50W load? My intuition says it's not the different output transistors, but that there is something else wrong. You might still have a dud transistor, sometimes an amp has enough muscle and feedback to keep on running even though one leg is stuck in class A. I would 'scope all the emitter resistors (not a multimeter) and look for the black sheep.
Also you could see who's putting out so much extra heat- 250W is a lot of quiescent power. Are you satisfied with 80Vrms output and +/-135V rails.
Also you could see who's putting out so much extra heat- 250W is a lot of quiescent power. Are you satisfied with 80Vrms output and +/-135V rails.
All of the output and switch transistors, 22 in all were tested and matched before being put into the amp. I use a regulated power supply running at the rail voltage of the amp in series with a 1 ohm resistor and the DUT similar in approach to what both pass and Elliot have published. I bias the first device to the whatever bias current the amp runs at and then sort through all the devices to find the closest matches. I then increase the test current and sort again. So all of the transistors were run at rail voltage and up to 100ma(the limit of my supply) each before being installed.
So i feel reasonable confident that the devices are in good shape. Crest/Peavey says they don't match devices at all. that the 21194 devices are "close enough" out of the bag. I wanted to do a better job of that and while my test method isn't perfect...at least its better then just grabbing a hand full and tossing them in the amp.
And I did check across all the emitter resistors with a DVM (not a scope) at idle and saw that all were idling about the same. none were running hot or drawing excessive current. I did not get a chance to do the same test with the amp running which may have proved to have been interesting.
and there may have been issues in the driver circuits as those were stock. I am wondering if these new devices don't have more base current draw??? I see that the spec sheets list a 5amp base current for the 21194 vs 1.5amp base current for the c3821's..could it be that the new devices need a lot more drive current and that's the extra current draw??
the 2amp idle current is normal for this amp, the fans run and all the housekeeping stuff. that's pretty normal. and the amp is rated at 700wpch at 8 ohms both channels running so one channel running 80vrms into 8 ohms is 800watts so I am pretty satisfied with that. these amps were built to be loaded down to 4 or 2 ohms and or bridged. My AC line in my shop will only allow for brief power testing at those power levels as I am in an office park running off one 15amp service. The new place will have better power!
I have an email into Crest/Peavey but trying to get any real real meaningful info out of them is difficult. If it dies this time, I will just send it in to the factory for repair, then when it comes back, open it up and inspect what they did and how they set it up.
I would LOVE to find some of the original Crest guys. engineers or repair techs that really know these amps. I did get in touch one guy, Lenny Tedesco but he only worked on some of the amps and not the CA or Pro series amps.
So i feel reasonable confident that the devices are in good shape. Crest/Peavey says they don't match devices at all. that the 21194 devices are "close enough" out of the bag. I wanted to do a better job of that and while my test method isn't perfect...at least its better then just grabbing a hand full and tossing them in the amp.
And I did check across all the emitter resistors with a DVM (not a scope) at idle and saw that all were idling about the same. none were running hot or drawing excessive current. I did not get a chance to do the same test with the amp running which may have proved to have been interesting.
and there may have been issues in the driver circuits as those were stock. I am wondering if these new devices don't have more base current draw??? I see that the spec sheets list a 5amp base current for the 21194 vs 1.5amp base current for the c3821's..could it be that the new devices need a lot more drive current and that's the extra current draw??
the 2amp idle current is normal for this amp, the fans run and all the housekeeping stuff. that's pretty normal. and the amp is rated at 700wpch at 8 ohms both channels running so one channel running 80vrms into 8 ohms is 800watts so I am pretty satisfied with that. these amps were built to be loaded down to 4 or 2 ohms and or bridged. My AC line in my shop will only allow for brief power testing at those power levels as I am in an office park running off one 15amp service. The new place will have better power!
I have an email into Crest/Peavey but trying to get any real real meaningful info out of them is difficult. If it dies this time, I will just send it in to the factory for repair, then when it comes back, open it up and inspect what they did and how they set it up.
I would LOVE to find some of the original Crest guys. engineers or repair techs that really know these amps. I did get in touch one guy, Lenny Tedesco but he only worked on some of the amps and not the CA or Pro series amps.
Zerocool,
Have you determined that the stock channel is drawing the proper design current. The problem may be in the unit that was not upgraded or changed and this may be operating correctly.
Just a thought.
Tad
Have you determined that the stock channel is drawing the proper design current. The problem may be in the unit that was not upgraded or changed and this may be operating correctly.
Just a thought.
Tad
That is a very real possibility! I left the stock channel alone for the most part. and from what i have seen before the stock channel seems to draw a more normal amount of line current.
then at this case cliping will be assymetrical ...either from rail to rail or from chanel to chanel ...
How did you adjust the bias? For a specific voltage across one emitter resistor or with a series resistor on the supply?
Clearly the current has to go somewhere, and there are only so many paths it can take. If you measure the voltage across the emitter resistors of the "inner" transistors and add all these up and then measure across R104 you've pretty much measured everything that should draw any significant current. If the sum of all these is much less than the measured total then something is definitely fishy, but with the voltage and currents involved this something should get very hot!
It's quite possible that the new transistors have a lower enough VBE that the idle current will increase significantly if the bias adjustment is left alone. I'd start there, and also scope the output for oscillations.
Clearly the current has to go somewhere, and there are only so many paths it can take. If you measure the voltage across the emitter resistors of the "inner" transistors and add all these up and then measure across R104 you've pretty much measured everything that should draw any significant current. If the sum of all these is much less than the measured total then something is definitely fishy, but with the voltage and currents involved this something should get very hot!
It's quite possible that the new transistors have a lower enough VBE that the idle current will increase significantly if the bias adjustment is left alone. I'd start there, and also scope the output for oscillations.
- Status
- Not open for further replies.