Crest CPX 1500 Power Amp - Noisy Fan

[TomCarlos]

I believe there is an issue with excessive current. Please advise if I should open a new thread. For now, this is what I can report.

Using a Kill-A-Watt meter, I see .5 amps when I turn the unit on. The current then slowly rises up to 2 amps at idle and hovers there when the fan kicks into high gear.

The thermistors are mounted at the tail end of the heatsink, at the hottest point.

I see the bias adjust pots, VR101 for channel a, VR201 for channel B. But honestly, I am not sure where to measure the bias current and not sure of the value. I can tell you that CR112 (right across the VR101) has .3 volts across it and the same for CR212/

[Leo_Gnardo]

I believe there is an issue with excessive current. Please advise if I should open a new thread. For now, this is what I can report.

Using a Kill-A-Watt meter, I see .5 amps when I turn the unit on. The current then slowly rises up to 2 amps at idle and hovers there when the fan kicks into high gear.

The thermistors are mounted at the tail end of the heatsink, at the hottest point.

I see the bias adjust pots, VR101 for channel a, VR201 for channel B. But honestly, I am not sure where to measure the bias current and not sure of the value. I can tell you that CR112 (right across the VR101) has .3 volts across it and the same for CR212/

You can start by marking the positions of the bias pots, so you can put 'em back to the starting point if necessary. Got a scope, right? Load resistors? Sine wave generator? OK, while monitoring your Crest's power draw with the handy Kill-A-Watt, send a sine wave into your Crest while observing outputs driving load resistors, preferably a low resistance say 4 ohms. You don't need to run high power, just so the amp's turning out say a quarter watt or so, and you see a nice clear scope trace. Turn one of the bias pots just a tiny bit & you'll see the current on the Kill-A-Watt rise or fall. Pick the direction that reduces bias current, dial it down a noticable amount say half an amp less AC current & watch for crossover distortion on the scope. None so far? OK then do the same for the other channel while observing its waveform for a crossover pip or notch. Still none? Kool! If that's the case you've reduced the bias current therefore heat by a good margin and maybe out of the zone where it will set off the hi speed fan switch. If you do see a crossover pip or notch, dial the bias up just a scoche until it disappears, then one tiny dab more & park the pot there.

Whew! Let your Crest sit there running for say 15 to 30 minutes, no signal, monitor the Kill-A-Watt, see it doesn't sneak up any more. If the fan doesn't go to hi speed, and you still have room to go downwards on the bias it's up to you to decide if you want to continue reducing bias 'til you hit the crossover notch point. Also, if the amp's rated for 2 ohms and you really think you're going to run it that low then you'd better test it with 2 ohm loads.

I'm not much in favor of running that low a load on any amp*, because although the amp may successfully drive resistors, some speakers have impedance dips at certain frequencies, typically some low frequency between 30 and 200 Hz, where the impedance dips below the speaker's nominal rating. You see where we're going with this? Let's say your PA has four 8-ohm woofers driven by one channel of your prize, 2 ohms right?, and the bass player starts standing on the note whose fundamental is the magic frequency that bottoms out the impedance curve. Whoompa whoompa whoompa whoo ... nothin'! Poof goes your amp and you're left wondering why. Well now you know.

*Asterisk for the rare exception, McIntosh amps with output autoformers that allow them to drive a wide variety of load impedances to full power. With them, you can get away driving low impedances as long as you're not whamming the amp into significant clipping. For that kind of toughness, you pay a lot extra, in cash and mass. Plus you have to do your arithmetic and use the right tap, it's not just plug & play.

[TomCarlos]

Ok Leo ... I will give this a try. But there has to be a place in the circuit where you can measure the bias current, right?

JPB mentioned the 60 degrees C before the fan kicks on...

I have a Tenma Desktop True RMS meter that has a crappy Type K Bead temp sensor. The top of the heat sink on Channel A is running hotter compared to the heat sink on Channel B. I was measuring around 47 degrees C on the Channel A heat sink and 40 degrees C on Channel B. And then... the fan kicks on to the higher speed.

One other annoying thing... I am looking at the Thermal Mgt circuit. It shows two thermistors per channel? I see markings on the PC board for 318A and 319A. But it looks like each heat sink has 3 thermistors. And I found R320 to R323. But for the life of me, I cannot find CR312 to CR315. They gotta be buried somewhere!

[Leo_Gnardo]

By monitoring the AC line current, you are indirectly observing the amp's bias current. OK, you're not seeing how many amps/milliamps each channel is drawing from the rails, but you would have to disconnect the amps from the filter caps, clip in ammeters that could stand a helluva lot of current if there was an unexpected current jolt, and reconnect them all when done. What a hassle! Plus one false move and poof. Or BLAM! There goes your amp and/or meter. It sure would be nice if some manufacturer put current sensing resistors in like we do for output tubes but I've yet to see it happen.

For temp measurement I got a $25 (at the time) laser thermometer from Parts Express. Works great on sets of output tubes - I can see which are running hot and which cool. It's a little harder with solid snake amps, the heat sink tends to average the temperatures of multiple transistors, but I can detect and monitor hot resistors & transistors. Heck I can stick it in my ear and take my own temperature that way too, what a hoot. Lots of fun & useful info for the price of a couple cheap lunches. And if you have a cat you can entertain it too I guess.

Dial down your bias pots, you'll have all those sensors running cooler.

[TomCarlos]

Got it Leo... that all makes sense. Thank you.

The amp seems to be in a better place now. I made note of the bias pot's positions. Both were the mid-center point. I used some DeoxIT F5 to clean them before testing.

For Channel A, I started with a 1Khz signal, 8 ohm, .5 watt output. Turning the Bias pot counter clockwise (to about the 1/4 position) decreased the overall current. BUT, I did not see any degradation of the signal, at .5 watt output, 1 watt output, 10 watts output. I then turned the pot to the full counter clockwise. Again, a reduction in current with no change to the signal output or a detection of a notch. I repeated that process with Channel B. Same results. So both pots are at full counter clockwise, max current reduction, and no apparent loss of signal at the output.

The amp has been running at idle for 30 minutes and is now pulling .33 amps AC. I'm gonna leave it running for another 30 minutes. If the same results, tomorrow I will retest this amp with some recorded music for an hour at a moderate volume level and see what the fan does !

And I found an Infrared Temp Sensor online- bought me one of those. Should be a cool tool to have.

[Jazz P Bass]

Well now .33 amps (33 watts) at idle seems a tad cold.

You could try the 'Enzo Method'.
Turn the bias pot until the amp starts to draw more current and then back it off a bit.

[TomCarlos]

I believe I found my answer on a thread posted at DYI Audio. Measuring the emitter resistors at the output (.33 ohms), we set the bias adjustment to give us ~7mv. That gives us around 21ma through the output transistors. According to the other post, that oughta do it for this amp. The total draw is now .76 amps AC. So we are getting closer to a more realistic setting.

[Leo_Gnardo]

I believe I found my answer on a thread posted at DYI Audio. Measuring the emitter resistors at the output (.33 ohms), we set the bias adjustment to give us ~7mv. That gives us around 21ma through the output transistors. According to the other post, that oughta do it for this amp. The total draw is now .76 amps AC. So we are getting closer to a more realistic setting.

OK, sounds good. Bias trims set exactly half way from factory, makes ya wonder, do they bother setting bias? Or just wing it...

[Enzo]

I have seen from Carvin and from Yamaha at various times, they used the mains draw. Similar to my cheap and easy method. They determined that when the amp is biased up to their spec, the mains draw will be some number. So if you set the bias control for that mains current reading, it will wind up biased correct. Sounds like winging it, but really it is legit.

[Jazz P Bass]

I am working on a Luxman R1120 (discrete output), & curiously, they specify setting the idle bias by measuring the current draw through the B- supply.

[TomCarlos]

Thanks Mates.....

I will review the replies and take notes for the future. Going back to a Leo comment about testing in different temps and thinking about a bias control for parallel output devices, I observed the following:

- I realize you can never dial in the exact same current through output transistors. So when measuring the voltage drop across the .33 ohm resistors, there will be a difference.
- Bias WILL change with temps. Setting the bias when the amp is WARM will not yield the same results when measuring the bias early in the morning when the room and ambient temp is 60 degree F. Last night, I was at .76 amps AC. This morning, when I turned the amp on, I started at .38 amps AC. Twenty minutes later, the current creaped up to .62 amps. The average voltage reading across the emitter resistors was around 1.5mv. That increased over the next 20 minutes.
- Bias drift can occur. Of course it can- it is temp related. In my case, when testing the amp last night, I tried two things. I chose a set of transistors from Channel A and the same position transistors in Channel B. Check the voltages on the emitter resistors with the chassis closed then open. The bias in Channel A seemed fairly steady. I did not see any significant change in the voltages on the emitter resistors. On Channel B, I could see a sway in voltage. I was monitoring the voltage on 2 different emitter resistors. Under a closed chassis condition (mini probes connected across the resistors and the top chassis cover temp placed over the amp) I was measuring around 7mv dc. When removing the cover, those voltages dropped down to 5mv. Is that a big deal? Eh, you tell me. I am thinking the thermistors in Channel B are reacting differently to changes in temp compared to those of Channel A.

To clarify something I posted earlier - I thought I was seeing multiple thermistors. My mistake. On Channel B, I see R319A. That device is at the far right of the heat sink (see photo). The other devices that are mounted against the heat sink are CR207 and CR210 (see photo, lines on the left and center). Those devices can be found on the Amp B section of the schematic, part of the diode stack. I cannot find an R319B device (as shown on the schematic). Perhaps there is no such device.

For learning, I am going to take some measurements over the course of the day and will hook the amp up for a real world live test with music at a moderate level. I'll report back.

Tom

[g1]

R319B should be on the other channel, also another set (2 blocks) of heatsink mounted diodes.
The thermistors should only affect fan speed and thermal shutdown, the diode blocks should give temperature compensation for the bias.
Having the lid on is important for any temp. related measurements or settings, it supplies the 'top' of the heatsink vent tunnel.

[JohnvL]

Hi Tom,

I found this on internet you have a Crest CPX model. I have a question if you can look at resistor R338. Its located and the power supply and next to the LM (head sink) thats run the fan (normal speed at 12 volts) I need the color code or value.

Its burns and destroyed. I have a CPX 900, but i think its the same value of the 1500.

(I work on a repair of a total fail of channel A. All transistors (208,9,10,12,13,14) diode (CR225), and drivers are blown. But try first run channels B. Also the 7815/7915 (+15 & -15v) are good at the 7815/79 but missing the +15 at the P108 connector that runs to the 2nd PCB (Ch A).)

thank you in advance

John (from the Netherlands)

[Jazz P Bass]

The schematic indicates that R338 is a 220 ohm 1/4 watt resistor.
It comes off of the 28Vdc supply & feeds IC 303 pin 8 (power pin)

Crest_cpx-900_R338.pdf

Make sure to check that CR321 zener diode is not shorted.