After you measure/record the outputs betas I think you could install one device per rail to make sure everything is functional. Should be $wise in case of disaster.
Depending on circuit RE and bias current, betas s/b around 10% for good current sharing. But maybe save the tightest grouping for the quasi pnp side. YMMV
Depending on circuit RE and bias current, betas s/b around 10% for good current sharing. But maybe save the tightest grouping for the quasi pnp side. YMMV
I'll second Infinia's advice of a single pair of outputs to start. You shouldn't need to remove the transistors, you just have to desolder 2 wires to test them still mounted on the heatsink with a DMM. You don't need to test them for beta, just use the diode check mode on the DMM to verify you get about a 0.7 volt drop from B to E and nothing with the other polarity and no shorts from C to E.
The fear isn't so much they were bad from Digi Key, but that your attempts to power up the amp, even with the light bulb, may have killed one or more of them. Generally if you killed an output transistor it will be be fairly obvious. It will likely read as a dead short on the DMM.
And stacked Large Advents are kinda cool! So I can understand your desire to keep the amp of a similar vintage. That speaker was legendary in its day. Sort of like the Phase Linear amps, it set a new price/performance point that took anyone else a while to beat. It's off topic, but do they by some miracle still have the original orange tweeters with the masonite (wood) bezel? The crossover frequency was really low so the tweeters often had a short life.
I also have to confess to owning a Phase 400-II that someone gave to me that's been boxed up in storage for many years. It has a weird random noise in one channel but otherwise was working when I boxed it up and, I think, is mostly original except for the op-amp resistors and a few replacement parts. It's a little beat up but not in too bad of shape aesthetically as I recall. So if you want to buy it for parts, let me know
Given what I wrote above, I've never bothered to invest any time in it beyond spending 15 minutes looking for the source of the weird noise. It would be kinda cool to have it (or parts from it) end up in a vintage application like yours. I've been meaning to put it on Craigslist. Shipping it to the other coast, however, won't be cheap.
EDIT: And I should add you probably don't need to remove any transistors from the driver board for testing if you can verify the input and VAS stage seem to be working OK without the output transistors hooked up. If the feedback loop can stay closed, the amp should go to zero volts out even with the outputs disconnected. Did you by chance check any voltages when you powered it up without the output transistors? Have you tried running it with just the drivers? If you care to post your schematic(s), or a link, that would help as I remember some of the design but not all of it. I may even have a service manual for the 400-II boxed up around here but that would involve a major search of my less than organized "archives"
The fear isn't so much they were bad from Digi Key, but that your attempts to power up the amp, even with the light bulb, may have killed one or more of them. Generally if you killed an output transistor it will be be fairly obvious. It will likely read as a dead short on the DMM.
And stacked Large Advents are kinda cool! So I can understand your desire to keep the amp of a similar vintage. That speaker was legendary in its day. Sort of like the Phase Linear amps, it set a new price/performance point that took anyone else a while to beat. It's off topic, but do they by some miracle still have the original orange tweeters with the masonite (wood) bezel? The crossover frequency was really low so the tweeters often had a short life.
I also have to confess to owning a Phase 400-II that someone gave to me that's been boxed up in storage for many years. It has a weird random noise in one channel but otherwise was working when I boxed it up and, I think, is mostly original except for the op-amp resistors and a few replacement parts. It's a little beat up but not in too bad of shape aesthetically as I recall. So if you want to buy it for parts, let me know
Given what I wrote above, I've never bothered to invest any time in it beyond spending 15 minutes looking for the source of the weird noise. It would be kinda cool to have it (or parts from it) end up in a vintage application like yours. I've been meaning to put it on Craigslist. Shipping it to the other coast, however, won't be cheap.EDIT: And I should add you probably don't need to remove any transistors from the driver board for testing if you can verify the input and VAS stage seem to be working OK without the output transistors hooked up. If the feedback loop can stay closed, the amp should go to zero volts out even with the outputs disconnected. Did you by chance check any voltages when you powered it up without the output transistors? Have you tried running it with just the drivers? If you care to post your schematic(s), or a link, that would help as I remember some of the design but not all of it. I may even have a service manual for the 400-II boxed up around here but that would involve a major search of my less than organized "archives"
All four still have the original fried egg tweeters and masonite on the woofers. I picked one pair up off Craigslist for $100 that were in perfect condition, and another for $75 that needed new surrounds. All four are in the bull nose walnut cabinets, although I actually prefer the look of the plainer "utility" cabinets. I bought caps to redo the crossovers, but I haven't got around to it yet. They sound great as it is, but I can tell my Marantz 2250b doesn't have the "umph" to really power them.
Just for fun, I tried turning the bias to zero with all 4 drivers inserted and just one output transistor. It powers up! Both LEDs jump a little on power up like normal, then settle to zero. Then the left channel LED jumps up about half way and starts dancing around a little. I tried adding a second output transistor, but it refused to power. I'm testing the transistors now.
Just for fun, I tried turning the bias to zero with all 4 drivers inserted and just one output transistor. It powers up! Both LEDs jump a little on power up like normal, then settle to zero. Then the left channel LED jumps up about half way and starts dancing around a little. I tried adding a second output transistor, but it refused to power. I'm testing the transistors now.
RocketScientist said:
The reason I mentioned beta was that I blew one of my amps and after removing all the transistors i found one in the LTP stage with a beta of one which was clearly faulty. This was a transistor that would plug straight into my multimeter to test the beta.
I understand its not as easy to do a beta test on the output transistors.
Its more about talking a logical approach to the testing procedure.
The more faults you can catch by being logical the better chance of ending up with a properly functioning amplifier.
blue lander said:
Yikes. Be careful. Time to get your old scope out and hooked up I think! That sounds A LOT like it's probably oscillating (or there's a weird intermittent solder connection problem or some such). As I explained above, it *might* be related to the light bulb but more likely you have some other problem. Is the light bulb changing in brightness with the "dancing"? Also, make sure the scope is the ONLY thing hooked to the amp. Leave the inputs disconnected and make sure it's not grounded in any other way. Otherwise the grounded scope can form a ground loop that can make the amp oscillate and possibly self destruct.
If you happen to have anything like a 1K - 5K resistor laying around you can hook up a speaker to the amp with the 1K - 5K in series and listen to what it's doing. But if it's a high freq. oscillation it will likely be well above 20 KHZ. If it's "motor boating" or some other problem you can probably hear it. But the scope will show you all of the above easily.
nigelwright7557 said:
Anything is possible I suppose, but with new transistors it's 99.9% likely they're either fine or obviously toast. But it comes down to how much time you want to spend. And it's also all too easy to damage PCB traces, and sometimes even parts, with excessive soldering/desoldering/soldering--especially on a well baked 30 year old single sided PCBs like Phase Linear used.
I'm no expert on Phase Linears but from what I can reckon from reading threads on DIYA (& not counting the reported loop instability) lots of the failures from either the factory or non expert repairs, are due to SOA failures from hard driven low/non sharing outputs. Given the low value of RE and low DC bias/small heat sinks. I don't know what the build procedure was at the PL factory, but maybe for the most part betas are close enough on new, same date code transistors.
I fired up the scope and the left channel is indeed oscillating while the right channel's dead quiet. I then paused to admire the workmanship of my oscilloscope. It's probably 15 years older than the PL400 and uses dozens of vacuum tubes, and it still works flawlessly!
I'm not sure what to make of the readings off my DMM, I get about .57v from B to E, but I get a 1.3volts if I do it in reverse. Shouldn't I get nothing? I think I need a better multimeter...
Now the thing won't power up even with the one output transistor that worked fine before. So I think I'm definitely damaging parts. I'm back to just the 4 drivers. Where should I be testing the voltage? Schematics for the PL400 model 2 are at http://home.earthlink.net/~dimlay/sitebuildercontent/sitebuilderfiles/400II-700IIschematic.pdf
EDIT: Using the old 2n3055 transistors I pulled out of the unit in the first place, I've almost completely repopulated the right channel. Some of them were bad and caused a short, but I *might* have enough to finished the channel...
I'm not sure what to make of the readings off my DMM, I get about .57v from B to E, but I get a 1.3volts if I do it in reverse. Shouldn't I get nothing? I think I need a better multimeter...
Now the thing won't power up even with the one output transistor that worked fine before. So I think I'm definitely damaging parts. I'm back to just the 4 drivers. Where should I be testing the voltage? Schematics for the PL400 model 2 are at http://home.earthlink.net/~dimlay/sitebuildercontent/sitebuilderfiles/400II-700IIschematic.pdf
EDIT: Using the old 2n3055 transistors I pulled out of the unit in the first place, I've almost completely repopulated the right channel. Some of them were bad and caused a short, but I *might* have enough to finished the channel...
Yes, Tek once made very nice scopes. Most considered them the best in the world. Today I think Agilent does a better job. But, regardless, even today's high-end scopes like my cost-more-than-a-new-car-Agilent, are plastic wonders that will probably be dead from lead-free surface mount solder problems in 10 years.
So is the 400 "happy" with Q107 and Q108 connected in the bad channel but not with any of the Q109 - Q114 pairs? If so, check the pair that used to work (i.e. Q109/Q110) with the DMM. And you said "adding a second output transistor" as in one transistor versus one *pair*. Are you connecting them in pairs (i.e. one positive aka Q110 and one negative aka Q109)? If not, you need to.
If you have the transistor pinout correct (beware the datasheet may show the pins from the *other* side view), the base to collector should read "open" in one direction and be a diode drop in the other. But that's if your DMM has a real diode range that reads voltage rather than ohms. Some cheapo meters just use less than 0.6 volts when in regular resistance mode and over 0.7 volts in the diode mode but still display resistance. Put another way, does the display have an "omega" (ohms) or "v" when it's in diode mode?
And I agree with Infinia's point about making sure the output transistors are halfway matched before getting serious about using the amp. He's right that if they're wildly different one transistor could "hog" most of the current and hence have a short lifespan. But that is only an issue for driving a real load at serious output levels. Right now you have other problems...
So is the 400 "happy" with Q107 and Q108 connected in the bad channel but not with any of the Q109 - Q114 pairs? If so, check the pair that used to work (i.e. Q109/Q110) with the DMM. And you said "adding a second output transistor" as in one transistor versus one *pair*. Are you connecting them in pairs (i.e. one positive aka Q110 and one negative aka Q109)? If not, you need to.
If you have the transistor pinout correct (beware the datasheet may show the pins from the *other* side view), the base to collector should read "open" in one direction and be a diode drop in the other. But that's if your DMM has a real diode range that reads voltage rather than ohms. Some cheapo meters just use less than 0.6 volts when in regular resistance mode and over 0.7 volts in the diode mode but still display resistance. Put another way, does the display have an "omega" (ohms) or "v" when it's in diode mode?
And I agree with Infinia's point about making sure the output transistors are halfway matched before getting serious about using the amp. He's right that if they're wildly different one transistor could "hog" most of the current and hence have a short lifespan. But that is only an issue for driving a real load at serious output levels. Right now you have other problems...
I'm not sure what the procedure is for closing the loop with just the drivers. But what were the betas of the outputs and drivers after this last "failure"? usually when an output goes it takes the driver and another stage or two. Don't be in a rush for these repairs or you'll end up 2 or 3 steps back.
RocketScientist said:
It's consistantly happy with Q107 and Q108 connected. If only those are connected in both channels, it always works (although the bad channel oscillates). It's when I start adding Q109 through Q114 that it gets starts getting angry. I actually managed to populate 109 through 114 with the old transistors. It ran for about 2 seconds. Then I heard a very quiet pop and the bulb goes bright again. I couldn't get it to power up until I removed everything from 109 to 114. I think I'm definitely burning up transistors. I'm going to see if Radio Shack sells a DMM that can read Beta tomorrow, so I can get better readings off these transistors.
infinia said:
I don't know about the 400-II, but I know with some Phase Linear amps, they stamped a code onto the output transistors so they could be halfway matched from the factory. I vaguely remember a tech-note/revision with the advisory that you purchase a full channel set from the factory if even one output device needed replacement. But I don't remember which Phase Linear amp(s) that was for.
I wouldn't, however, call 0.33 ohms a "low value of Re". Some, like Douglas Self, claim you rarely need anything greater than a third of that value (0.10 ohms) without any special matching required. Most of my own DIY designs have used MOSFETs or a single pair of BJTs so I haven't done much research into if Self is correct or not. I do know it's very worthwhile to match MOSFET's in multi-pair designs.
blue lander said:
Ouch. You might try a lower wattage light bulb. Or even better, see if Radio Shack has some 5 watt or 10 watt 10, 47 (50) or 100 ohm power resistors. If so, try temporarily soldering those around the F2 and F3 DC rail fuse holders and remove the fuses. Although I don't see any caps after the fuse holders which is another Bob Carver blunder. So see if they also have some 0.1uF 100V or higher film (i.e. polyester or even ceramic) capacitors and solder those across the fuse holders (in parallel with the power resistors) to help maintain a low AC impedance at high frequencies so you hopefully don't add to any oscillation problems.
The problem with the light bulb trick (or even a variac) is if the amp does anything "bad" the full charge of the filter caps can be dumped into the transistors. And while the long term current is limited by the bulb, the short term current can still be hundreds of amps which is likely what's destroying your transistors. The 10 - 100 ohm resistors will limit the peak current to something the transistors can handle.
The other thing, if you're not doing it already, is try to power up the amp as briefly as possible *watching the scope* and immediately kill the power if you see any oscillation. It would be very useful to know if oscillation is what's killing devices, or you're have some other problem. If it just dies with no bursts of oscillation that's a very different symptom.
someone could check me.. you have MJ21196 I think these have over twice the SOA and 4x ft of the original outputs going from memory. So you could get by with 2 pairs instead of 3 FWIW. Stability might be a problem tho.
As someone else said ...you could close the loop w/o the outputs and drivers that can save a lot of grief here. You might have to up the Vas miller cap value for stability.
As someone else said ...you could close the loop w/o the outputs and drivers that can save a lot of grief here. You might have to up the Vas miller cap value for stability.
Yeah, I'd wait to do the resistor in place of the fuses trick above which should at least give you time to turn the amp off when it starts to oscillate. I should add if you do that don't try to drive a real load past a whisper with the resistors in place and don't crank up the bias too high.
While you're at Radio Shack (or ordering from DigiKey or whatever) also try to get a few 100pf - 470pf ceramic caps to temporarily add one across (in parallel with) C113 on the bad channel. That's the compensation cap Infinia is referring to (known as "Cdom") and should make the amp more stable at the expense of slew rate and HF bandwidth.
The other compensation cap is C106. And if the above fails, you might also try adding a 100+pf across it. The fact they're not both 100pf or 120pf, however, tells me the original compensation was probably tricky. Manufactures like to re-use the same parts whenever possible. So the fact they couldn't both be the same value is a clue the design may be marginally stable.
Unless I'm missing something the loop *IS* closed with the just the drivers (Q107/Q08) connected and the amp is perfectly stable then. You should still have the scope connected, however, with the outputs disconnected and verify it's not oscillating or slammed into a rail. You could even try an input, ideally from a portable source like an ipod (with no ground), and see if the output looks OK on the scope. The driver stage should drive just the scope perfectly with perhaps some crossover distortion due to bias issues.
I'm also not sure how the amp is physically wired, but the wiring from the transistors to the driver board should be kept as short as possible and probably be twisted together for each "rail" of outputs. Oscillation happens from phase shift and positive feedback. And the wrong wiring can cause either or both.
It's also possible the amp needs *some* bias to be stable. So if you have the bias still at zero you might try increasing the pot, again with the outputs disconnected until you read a small voltage (no more than 0.1 volts) across R120 and also check for the same voltage across R130. That means the drivers are biased just slightly "on" with no more than 10 ma of bias.
If the above fails to kill the oscillation, making R127 and R129 smaller may also help. Perhaps 100 ohms instead of 150 ohms. So some 100 ohm resistors might also be valuable on the shopping list.
While you're at Radio Shack (or ordering from DigiKey or whatever) also try to get a few 100pf - 470pf ceramic caps to temporarily add one across (in parallel with) C113 on the bad channel. That's the compensation cap Infinia is referring to (known as "Cdom") and should make the amp more stable at the expense of slew rate and HF bandwidth.
The other compensation cap is C106. And if the above fails, you might also try adding a 100+pf across it. The fact they're not both 100pf or 120pf, however, tells me the original compensation was probably tricky. Manufactures like to re-use the same parts whenever possible. So the fact they couldn't both be the same value is a clue the design may be marginally stable.
Unless I'm missing something the loop *IS* closed with the just the drivers (Q107/Q08) connected and the amp is perfectly stable then. You should still have the scope connected, however, with the outputs disconnected and verify it's not oscillating or slammed into a rail. You could even try an input, ideally from a portable source like an ipod (with no ground), and see if the output looks OK on the scope. The driver stage should drive just the scope perfectly with perhaps some crossover distortion due to bias issues.
I'm also not sure how the amp is physically wired, but the wiring from the transistors to the driver board should be kept as short as possible and probably be twisted together for each "rail" of outputs. Oscillation happens from phase shift and positive feedback. And the wrong wiring can cause either or both.
It's also possible the amp needs *some* bias to be stable. So if you have the bias still at zero you might try increasing the pot, again with the outputs disconnected until you read a small voltage (no more than 0.1 volts) across R120 and also check for the same voltage across R130. That means the drivers are biased just slightly "on" with no more than 10 ma of bias.
If the above fails to kill the oscillation, making R127 and R129 smaller may also help. Perhaps 100 ohms instead of 150 ohms. So some 100 ohm resistors might also be valuable on the shopping list.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.