Peavey CS800S "Tempsistor" testing?

[stoneattic]

Hello all,

I'm trying to repair an old Peavey CS800S power amp and the symptoms match what I have read to be a faulty "tempsister" (that's Peavey's name for it), which is one channel working and the other not, but showing the DDT led and the fans running at full speed.

I emailed Peavey the serial number requesting schematics and they sent me the package below, which has schematics, parts lists, etc.:

CS800S COMPLETE.zip

The items in question are R206 and R207 on the Main Board schematic, which are called out as:

235102AF 70240216 1K TEMPSISTOR HAND-INSERT MULTI-ROTATE FENWAL

I pulled the tempsistors and measured them with my DMM. Initially the resistance is all over the place, from short to open, and then after awhile settles around 1K. Is this how they are supposed to measure? This is not the response my DMM usually has to normal resistors. The 1K makes me think they are okay, but I'm confused by the odd initial measurements. My understanding is that they are resistors that change resistance with temperature so I expect them to test like normal resistors.

Any experience, suggestions, etc., would be appreciated! Thanks in advance!

[nevetslab]

Are the thermistors still in circuit, or now removed? I assume these are either beads, packaged somehow with short thin wire leads that are soldered into the PCB in some fashion. Thermistors are available in a variety of values, but in this application, they are no doubt NTC (negative temperature coefficient) type, meaning as they are heated, their resistance decreases. So, if you're holding one while trying to get a stable reading, your finger heat is causing them to change. That's what they do. Sit them on the bench, connect your DMM to the leads of each, and let them stablize to the temperature environment. Then, to verify they are working, you can press your finger to them, which should cause the reading to decrease, or even apply heat from your heat gun or soldering iron....just barely, as that is outside their working temperature range (heat sink temperatures range from room temp to around 80-90 deg C).

If these thermistors have failed, you could place a resistance decade box between CR117 cathode and ground (they're using Chassis ground in this section of the schematic). I don't know if the fans run when the thermistors are at room temp or if they start up when the heat sink temp reaches 40 deg C. There is a small chart adjacent to C152 of the fan control circuit schematic, giving you several voltage levels that pertain to the temperature of the heat sink, and the resultant voltage applied to the van, so it definitely is a temp-variable speed control for the fans. You could use this to dial in 1k to start, and verify the fan changes speed as you decrease the resistance. Thermistors are readily available from vendors, both in bead form as well as some packaged.

I had just ordered 10k NTC bead thermistors for a few bucks off ebay, and a package of 10 pcs of aluminum #6-32 x 1/4” male-female stud-standoffs, and dropped in a drop of freshly mixed 2-part thermal epoxy into the standoff holes, then dropped the bead probes into it and let them cure overnight. Used to replace a missing thermistor on a BGW GTC Power Amp heat sink, and controls the variable temp fan speed just like your Peavey CS800.

[stoneattic]

Thank you for the reply nevetslab,

I tested them out of circuit. They look kind of like a 1N4148 diode with clear plastic tubing over the exposed lead:
CS800S tempsistor.jpg

I assume the tubing is to prevent it grounding out on the heat sinks since they are right up against the heat sinks. It didn't occur to me when testing that my touching them would affect them, but now it seems so obvious.... once you pointed it out.

Now I'm pretty sure they are fine but I'm thinking just to be sure while troubleshooting the amp I may replace them with regular 1K resistors. I don't believe that should cause a problem since I won't really be running the amp under any strain so fans speed shouldn't matter. If going to the resistors gets it working I will order new tempsistors from Peavey. If it doesn't, then I've eliminated a common cause.

I was really hoping they were bad since that would be a simple fix. I'm not looking forward to figuring out what the problem is. It might be beyond my very limited skill set. I'm better with tube amps and pedals. Solid state amps are kind of voodoo to me.

[Enzo]

Well they are not regular resistors. They are temperature reacting resistors. If you measured them right after removing them, they could still be hot from the soldering process. Give them time to cool down and they measure right.

Those thermistors are just one of several things that control the fans and lights. SO explore that whole mess of stuff down by Q109, Q110.

[The Dude]

If you have one of the DDT lights on, I'm more apt to suspect a problem with that amplifier circuit. FWIW: My first check on a CS800 is to measure across the speaker connections with a DVM set to diode check (of course with power off). If the amp has shorted outputs or triac, the meter will usually tell you. You'll sometimes see caps charging, but the meter should eventually go to an open circuit reading. If it doesn't, you know where to look.

[stoneattic]

I re-measured them this morning and they appear to be fine at 997 ohms. So I guess I'll reinstall them and start testing around Q109 and Q110 as Enzo suggested.

The nice thing is I have a functioning channel to compare readings on. The bummer is the CS800S (2U) has them stacked so I can only measure one at a time powered up, where as the old CS800 (4U) had them side by side and was convenient to work on (but not to carry). I guess I can do an in circuit comparison un-powered as a first step? I know it won't tell me everything but might point out something really amiss?

The Dude, does the test you suggested work for all solid state amps? As I said before I'm pretty new at this and it sounds like a great starting point.

Thanks all!

[nevetslab]

Thank you for the reply nevetslab,

I tested them out of circuit. They look kind of like a 1N4148 diode with clear plastic tubing over the exposed lead:
CS800S tempsistor.jpg

I assume the tubing is to prevent it grounding out on the heat sinks since they are right up against the heat sinks. It didn't occur to me when testing that my touching them would affect them, but now it seems so obvious.... once you pointed it out.

Now I'm pretty sure they are fine but I'm thinking just to be sure while troubleshooting the amp I may replace them with regular 1K resistors. I don't believe that should cause a problem since I won't really be running the amp under any strain so fans speed shouldn't matter. If going to the resistors gets it working I will order new tempsistors from Peavey. If it doesn't, then I've eliminated a common cause.

I was really hoping they were bad since that would be a simple fix. I'm not looking forward to figuring out what the problem is. It might be beyond my very limited skill set. I'm better with tube amps and pedals. Solid state amps are kind of voodoo to me.

I haven't seen that form factor for a thermistor before. You're right, it does look a lot like a 1N4148, though thicker leads, and no cathode marking.

Yeah, I can understand the feeling of being a bit overwhelmed by an amp like this, with SO MANY MORE PARTS than a tube amp to contend with, and all of them need to be operational so the current flow thru every stage works, and influences the next stage, all the way to finally tickling the loudspeaker properly. But, take heart...there are a lot of us here that will be glad to lend a hand and direct you, step by step as needed. Thankfully you have good documentation, and Peavey does a good job supporting their huge customer base.

[The Dude]

......The Dude, does the test you suggested work for all solid state amps?......

Nope, not for all amps. Some amps have a relay on the output that closes when the amp is powered up. In that case, the output connections are basically disconnected with the amp off.

[stoneattic]

So I put the tempsistors back in tonight and noticed that R184 and R175 looked burned. I pulled them and tested them and R184 (should have been 47) read open and R175 (should have been 150) read shorted so I replaced them both. R184 is supposed to be flame proof, which I didn't have laying around so I put a regular carbon film in. I assume this is okay for testing/troubleshooting and if I get it fixed I will order the flame proof.

I put the board back in and I am getting different results, unless I am mis-remembering (always a possibility).

- With the "bad" board A completely unhooked channel B works fine and the fans turn slow as expected.
- With the "bad" board A connected to the PS (and all other connections) I hear a relay click but no power anywhere, nothing comes on.
- If I disconnect the PS connections to the "bad" board but leave the others connected channel B works fine, the fans run fast, and the DDT light is on on the "bad" channel A and that channel is dead (no "on" light, no sound)

My recollection was that with the "bad" board A connected completely, including PS, channel B would work and the "bad" channel would show DDT and not come on. I'm wondering if I messed something up when I was doing my initial troubleshooting, which burnt the two resistors, which I don't remember being burnt before. Although if they cooked when I was under the hood I would have thought I would have noticed that.

Nothing else looks visually damaged. My next step is to pull both boards and start taking resistance measurements, starting around the burnt resistors looks for differences. It looks like both resistors are right off the -75V PS connection so it makes me wonder if I connected the PS wrong at some point.

Any thoughts (besides I may be an idiot), ideas, direction, etc.?

As always, many thanks!

[nevetslab]

Most likely both R184 and R175 burned due to excess current being pulled by failed semiconductors. R175 is at the bottom of the voltage gain stage that includes the bias diodes to set up the output stage. The output stage, beginning with Q124/Q125 pre-drivers, Q126/Q127 drivers and Q128, Q129, Q130, Q131 & Q132, Q133, Q134 & Q135 output xstrs follows the two drive signals, and just provides the current to drive loudspeakers from that voltage gain stage & the two stages preceding it.

I'm guessing you haven't become familiar with using your multimeter's diode test mode. Or, if it's an older multimeter not having one, you actually do. It's the 2k range in your ohmmeter. The diode test mode or the 2k ohmmeter range provides a 1mA constant current source, and on diode test modes, usually has a voltage source of around 3VDC. In the 2k Ohmmeter range, it's a 2VDC source. Good enough to test most semiconductors, though with GREEN LED's, you need more than 2V source to light it up (2.2VDC is typical for a GRN LED). But it's totally adequate to use for finding shorted semiconductor junctions on your NPN and PNP xstrs, signal diodes, rectifier diodes and zener diodes. Failure mode in solid state amps like your Peavey CS800S result in shorted transistors, shorted diodes, open junctions in transistors, shorted or open zener diodes and the like.

This is what The Dude was referring to, looking between ground and the output terminal of the amp to see if there's a short. But, with relays connecting the amp to the speaker, usually amps like this would have DC Fault Protection, and keep the relay open. This amp has both relay connection with it's control logic, and a Triac circuit to impose a dead short across the output in the event one or more output transistors fail in a short.

If testing xstrs and diodes is new to you with a multimeter, here are the basics. With the multimeter in diode test mode, red & black leads...the RED lead is the positive voltage source, the black lead is the return, and you are applying 1mA current thru the leads. Connecting the red lead to the Anode of the diode, and black lead to the cathode, you would typically read 0.665V or so....we can round it off to either 0.7V or 0.6V. That's a good semiconductor junction. With the leads reversed, you will measure open circuit. Testing an NPN transistor, connect the Red lead to the base, and black lead to the emitter, and you'll measure about the same 0.6 to 0.7VDC. Connecting the black lead to the collector would be the same. With a PNP transistor, you reverse the leads, with the black lead to the base, red lead to the emitter, then to the collector, and again, read about the same 0.6V to 0.7V. If measuring power transistors (like Q28 thru Q35), the reading will be lower, typically 0.51V to 0.56V, as there's more silicon in the semiconductor junctions.

Zener diodes will measure a normal junction between anode and cathode (0.6 to 0.7V), while their real function is establishing a voltage reference, as found on CR120 1N751A, which is a 5.1V zener diode.

Testing for shorts with everything connected (all soldered into the PCB, wired up, etc), you're looking for shorted semiconductor junctions on the diodes and transistors. I assume you know the symbols for NPN and PNP's. If not, the arrow pointing away from the base on the emitter is NPN, and if pointing toward the base, it's PNP.

So, your first real task is to find and list all of the shorted semi's. There will be some, as R175 & R184 burning are solid clues. I'd begin with the voltage gain stage consisting of Q120 PNP and Q121 NPN, and the diodes between them CR127 thru CR130, as well as the diodes connected between the base and collector of Q120 and Q121 CR126 and CR131.

If you find shorted pre-drivers, drivers and output xstrs, there are likely to be open emitter resistors that connect to the output buss that runs to the output relay thru an inductor & power resistor, ahead of the relay and triac circuit.

I haven't looked at all of the circuits in this schematic to see if the DC fault sensing circuits can turn off the power supply to the 'failed' power amp. This should be enough to get you started.

Also, since both amp channels are common to the +/- 75VDC power supply, your bad channel doesn't appear to be pulling it down and blowing fuses, so there's a good chance the damage is not in the output stages. But, you need to check them all. We can deal with all the rest of the front end circuits using op amps and such later. Look over Q114, Q115 and beyond at this point.

[stoneattic]

Wow nevetslab, thanks so much for this!

My DMM does have the diode test mode, but I've never used it for anything besides checking a diode out of circuit, and didn't realize I could test LEDs with it. I just used a 9v with a resistor to check them. That's going to save me a ton of time in the future.


The more I think about it to more I think I caused additional damage (burning R175 and R184). I'm close to 100% sure they were not cooked when I first opened it up. The only thing I can think of is I must have reconnected the PS incorrectly at one point, so there is probably more than one "issue". The original and the one I likely caused.

I will sit down this weekend and go over the entire board checking every diode and transistor. Hopefully they are readily available components, with the age of this guy I'm always a little worried about finding replacements.

Thanks again!

[nevetslab]

When I was looking at the schematic, I'm seeing a lot of in-house semiconductor device numbers, as Peavey is one of many who have their own part numbers silk-screened onto the parts. I do have a list that came from Peavey that gives a cross-reference from their P/N's to the industry numbers, like they use in the front end of the discrete power amp. We can cross that bridge after you find what's ailing this. Since R175 burnt, also check R172, its' symmetrical counterpart off the positive supply rail. Both pass the same current, so it may also have been affected, as well as that positive rail series resistor R176.

It looks like the Peavey in-house numbers begin with Q120 & Q121 upward. As for replacement parts that may be obsolete, I can help you out there with suitable parts that can be bought thru normal distrubuters like Mouser Electronics and such. Don't got the NTE route for replacements.

[stoneattic]

I finally had a chance to start really digging in last night/this morning. Here's what I've found so far, and I believe it doesn't look good:

I started with checking Q120 and Q121 and the diodes in between as nevetslab suggested using my DMM's diode test mode:

"bad" board
Q120 .463/.000
Q121 .366/.458

"good" board
Q120 .464/.368
Q121 .368/.454

I found a cross-reference for Peavey transistors (I assuming the same one enevetslab has) and they appear to cross to MPS-U 60 and MPS-U 10 (both obsolete). When I look up the datasheets they are shown as a different form factor so I'm not sure which is the collector or emitter, but based on at least the Q120 measurements compared to the good board, they need to be replaced.

CR126 .457/OL
CR127 OL/OL
CR128 OL/OL
CR129 .519/OL
CR130 .041/.041
CR131 .462/OL

The "good" board measured the same. CR127, 128, 129 where hard to find. They were in clear plastic tubes in the heat sinks. CR127 & 128 are dual diodes, which are new to me. I'm not sure if using the standard diode test setting is valid.

I checked a bunch of other transistors and diodes as well (not all). Some appeared to check out good while some others confused me. I expected either something around .6 or .7 but some would have a reading in that range between the base and emitter or collector, but the other leg would read ~.1 (Q122, Q123). A couple read .000 (short) on one leg (Q112, Q124) so I assume they are bad. I still need to compare all the rest to the "good" board.

Everything up to that point was tested in circuit. I decided to measure the big transistors (Q127 through Q135) since if they are failed they would likely be the most expensive to replace. Q128 though 135 appeared to be shorted in both directions. I compared what I was reading there with the "good" board and the "good" board transistors measured how I expected. When I measured CR140 and 141, since they appeared to "feed" those transistors, they tested as shorted on the "bad" board, but not the "good" board. I lifted one leg of each on the "bad" board and they then tested fine. I guessing that 129 through Q135 being shorted are creating a path around CR140 and 141?

Do I need to lift a leg of everything that tested bad and recheck? Any chance the Q128 through 135 are NOT dead?

So for the long babbling post but I'm trying to make sense of what I've seen so far.

[nevetslab]

I'm assuming your DMM semiconductor readings are in the form of "base-emitter/base-collector". I'm used to seeing somewhat higher junction readings, even when soldered in-circuit, and the probing order differs on Q120 & Q121 on the good board, but no matter. On the lower reading of those, you're probably seeing the effects of the resistors in parallel with the base-emitter junction, or the diode across the base-collector junction. Q120 is reading a shorted junction where you record .463/.000. I didn't realize these parts were TO-202 packages. Yes, those are obsolete. You can, though, replace them with MPSW-42 (NPN) and MPSW-92 (PNP), which is a 1W TO-92 plastic package, having the same pin-out mechanically. The schematic shows 3mA current running thru them, so with 73V across them, that's about 220mW. The smaller TO-92 1W package will run hotter than the TO-202 parts. To use a higher-wattage package, such as that of Q124/Q125, the pin-out is E-C-B left-to-right, so lead bending would be in order. MJE340 (NPN) and MJE350 (PNP) would be a suitable part, would run a bit cooler, it being a larger part.

With CR127 & CR128 measuring the same on both boards, I'll take that as they're good on your bad board. Not sure why a dual-series-diode wouldn't measure around 1.20 to 1.30V, but, no matter. Those didoes CR127-CR130 are the thermally-coupled bias diodes, which track the temperature of the heat sink and keep up with the thermal change on the output stage.

Q122 & Q123 are sensing the current thru the output stage emitter resistors, and as such, the base-emitter junction will read low, as you have 3 resistor networks of 1k + 0.33ohms in parallel, yielding about 334 ohms across the base-emitter junction of Q122/Q123.

Yes, Q112 and Q124 would be bad with that reading. I forgot to mention that in checking xstrs, to check across the collector-emitter as well. Now, in circuit, that will often give confusing readings if you don't understand the circuit and what else is connected. When you do find shorted junctions, it's wise to check that C-E junction, particularly on the output stage. Q124 is a pre-driver, and if it had C-E short, that will make Q126 appear to have a shorted B-C short.

Normally, on the output stage Q126 thru Q135, you should be reading good junctions between base and collector of all those. Base-emitter readings will appear low due to the emitter resistors all being common to the output buss, and give you very low readings.

When you lifted CR140 and CR141 back-lash diodes across the output stages/output buss, did Q128 thru Q135 measure ok? Now, with the outputs in-circuit, the base-emitter junctions will read extremely low, but not shorted. If they read the same as your 'good board', then you're ok. Takes a little getting used to spotting a shorted output xstr that's one of several in parallel, but, there are differences found, where one or more that are truly bad will read lower than the others. You're measuring the base-emitter junction with 0.33 ohms in series with 22 ohms across the base-emitter junctions. The 22 ohms is the emitter resistor to output buss from the driver, which is in parallel with the outputs' base-emitter junctions, each with 0.33 ohms in series with their emitters to the same output buss.

Were the backlash diodes CR140/CR141 ok?

Another clue to suggest the output xstrs are ok, is that both amplifiers are powered from the same power supply. If one of the power amps' output stage had shorted outputs, it would pull high current, drag the power supplies down. Unless there are series fuses to each power supply of each power amp board. I didn't notice there being any. Quick way to tell is measure between the collector 'cases' of each amp to see if it's common (short). If not, then there may be fuses that are open on the bad amp.

A further check, is, of course, to remove Q126 thru Q135 from the board, and check them out-of-circuit. When facing the bottom of the part, with the base & emitter resistors below the 'center-line', the terminal on the right is the base, terminal on the left is emitter, and collector, of course, is the case.

[stoneattic]

I took the readings from the base to emitter and base to collector, but I didn’t differentiate, so I can’t tell which order I listed the values in. I didn’t look up which were NPN vs PNP but tried it one way and if it didn’t measure I flipped the DMM leads. I did not measure across the emitter-collector. I assumed all of them, except the big TO-3s, had the base as the middle lead and the collector and emitter as the outside leads.

When I lifted CR140 and CR141 Q128 thru Q135 still test shorted. CR140 and CR141 tested good when lifted.

I wouldn’t assume that any component failures happened in operation. I was not the first person to work on this. Someone else had been in there before and I very well could have caused some of the damage myself in my first attempts to figure out what went wrong.
I’m thinking my next step is to pull the output xstrs and measure them out of circuit?

Any hints on how to do that? On smaller xstrs and other components (resistors, diode, caps, etc.), I usually use my Hakko FX-888D set to 750°F and Soldapult. Occasionally on small items, or really tight areas like ICs, I will use solder wick. I haven’t tried yet, but the TO-3s on big heat sinks look like they require a lot more heat. I have a couple of old high wattage irons and guns that seem like they may be more appropriate for these.