| ricsmuts |
I built this amp X2 (see attachment) as a two channel car amp
It works great, I've biased it to about 20ma
If I push the amp I keep having thermal runaway. Obvoiusly I've done something stupid somwhere...... and it's cost me a small fortune in output transistors. I've tried almost no bias as well (5ma)with the same result.
Is TR10 supposed to be attached to the heatsink?
Any ideas? |
|
|
| hitsware |
| I would surmise D1,D2, and D3 should be on the heatsink |
|
|
| Perry Babin |
The bias components (transistor and diodes) need to be attached to the heatsink. You may be able to mount only the transistor on the sink for proper compensation but having all of them should provide better compensation.
Insert 5 amp fuses in the positive and negative rails between the supply and the amplifier. This will allow you to run the amp hard enough to make it heat up but should protect the power supply and the outputs. Drive it as hard as you can without blowing the fuses. Every couple of minutes, remove the input signal and re-check the bias current. Do this until it gets as hot as you intend to run it. If you have thermal protection, continue to check the bias current until it reaches thermal shutdown. |
|
|
| ricsmuts |
Thanks Perry
Will try that.
I've realised again just how destructive thermal runaway can be. Fuses are essential. |
|
|
| luka |
Hi
I would even try without fuses, + since they are so cheap and effective |
|
|
| ricsmuts |
Is this circuit correct?
I have moved the transistors and diodes now, and it's a bit better.
What seems to be happening is both rails of output transistors turn on at high volume or high power transients, just below this level I can let the amp get very hot and it has no problem.
Only occurs with peaks |
|
|
| Perry Babin |
If the bias current holds at all normal operating temperatures but it spikes for a few seconds just after a period of very high power output, the bias components are not close enough to the output transistors. The temperature of the output transistors is increasing but the temperature of the heatsink where the bias components are mounted is significantly cooler than the output transistors. Move the bias components as close to the outputs as possible.
This can also happen when the insulator material between the transistor and the sink is inefficient. The transistor sees a spike in temperature and it takes longer than it should for it to pass the heat to the sink. As the transistor returns to the temperature of the heatsink, the bias current returns to normal. |
|
|
| ricsmuts |
Maybe I should those thermaltrak output transistors I have lying around.
It's awkward to move that whole circuit to the heatsink, what would you reccomend is the minimum components?
Thanks for the advice so far. |
|
|
| Perry Babin |
You should try heating the bias circuit components individually to see which one causes the greatest change in bias current. Heat the diodes as a group. If heating the transistor alone creates a significant change in bias current, try relocating only the transistor. If both the diodes and the transistor make a big difference in bias current and moving the transistor alone doesn't provide enough compensation, you'll have to move all of them.
If you mount them on a small piece of proto-board, they will be easier to relocate as a group. |
|
|
| ricsmuts |
I deserved that nice logical answer!
I could relocate most of it with two wires and proto board.
Once I get that right, I'd like to increase the VAS (not sure the drivers are being driven hard enough)
Once it's stable I will post the PCB I have done, which was designed to fit into a standard type car amp heatsink:D |
|
|
| luka |
Hi
| quote: | | fit into a standard type car amp heatsink |
Which is? there are standards for that?? |
|
|
| ricsmuts |
No, I mean the average size is often 200 -210mm inside for the heatsink.
The amp and SMPS are about 190 wide, so a borrowed heatsink can often be used.
Where can I buy proper car amp heatsink, cannot find any locally? |
|
|
| luka |
Hi
Nether do I, probably only if broken amp or something like that |
|
|
| ricsmuts |
Heated TR 9,10,11 and diodes separately and all that happens is the bias current goes up very slightly, I mean like 2mV
I think I'l put the diodes alone on the heatsink |
|
|
| ricsmuts |
| The amp and it's SMPS (1) |
|
|
| EWorkshop1708 |
I'd recommend the ThermalTraks if you have them......
And look around for big heatsinks of several kinds, you want good heatsinks to avoid thermal runaway!
I went to an Electronic Surplus outlet in my state to get big heatsinks, and its hit or miss with that, depends on what they may have on hand, but I managed to get some monster heatsinks. :D
See if a repair shop has some old car-amp heatsinks or try a surplus outlet. There's also a guy on Ebay that sells it by the foot, so you have to use what options are availiable to you to get good heatsinks. I've collected heatsinks, and used them years later because good ones are hard to find sometimes.
I'm using the ThermalTrak in my new design. This was because I didn't want the trouble of mounting extra parts for thermal compensation, and the built in diodes are convienient!
On this pic, the ThermalTraks are the main devices.
On another amp (subwoofer) I use the Thermaltraks as drivers to compensate for the temp of several pairs of outputs (all on the same heatsink) |
|
|
| ricsmuts |
| The amp and it's SMPS (2) |
|
|
| EWorkshop1708 |
| quote: | Originally posted by ricsmuts
The amp and it's SMPS (1) |
Holy **** that's a nice looking amp! :bigeyes: |
|
|
| ricsmuts |
WOW, you don't go easy on the heatsinks! I don't think you will have any heating problems there:D
If you ever consider posting that thermal track design, look me up please.... |
|
|
| ricsmuts |
Thank you:cool:
My attempt at brute force engineering.
Pity it's not thermally stable yet, need real heatsinks like you have lol |
|
|
| Perry Babin |
| quote: | | Heated TR 9,10,11 and diodes separately and all that happens is the bias current goes up very slightly |
TR10 is the only transistor used for bias compensation.
Something is really wrong if you heated the bias compensation components (TR10, D1, D2, D3) and the idle current INcreased. The idle current should DEcrease when those components are heated. |
|
|
| luka |
Hi
WoooooW that is big, snap !! |
|
|
| ricsmuts |
The schematic in this post is the one I used, the dots where the wires join are very vague, so a mistake could have been made.
(any designers out there care to redraw the dots...pretty please)
It sounds great at medium vol, no distortion on the scope.
Where oh where do I start to sort this out, I'm not a beginner, but don't have ton's of experience either.... |
|
|
| Perry Babin |
The red dots appear to be connections. Is this what you have?
Before making any mods, install fuses. |
|
|
| ricsmuts |
Yes, those I have connected as shown.
I will try heat TR10 totally on it's own, or because it's mostly for a sub, just convert it to pure class B, but I feel that would be a waste
I was not sure about between TR13 and TR14, are they only connected together with the 180 Ohm resistor, and to the base of the output transistors?
That part of the circuit looked odd to me |
|
|
| Perry Babin |
| If you meant 13 and 15, yes. |
|
|
| ricsmuts |
Thanks for all the advice Perry
I am going to double check the circuit and values step by step, move TR10 on it's own, and get some heavier heatsink.
I would love to get this sorted as it needs to sound as good as it looks.
If I can get it all working, will post all the PCB's as it's a nice project for a sub amp. |
|
|
| 81bas |
Hi!
At bottom of the schematic is some note present: TR10 HARUS DIPASANG BERSAMA PENDINGIN POWER TRANSISTOR... I am not able to translate this, but I think it is saying, that at least TR10 should be on the heatsink with the power transistors together :D Diodes alone are not enough, though...
Regards:rolleyes: |
|
|
| ricsmuts |
Well found it
The language is Bahasa (Malaysia) Translated it is "SHOULD FIX ALL POWER'S COOLER TRANSISTOR "
No what on earth does that mean?:xeye: |
|
|
| luka |
Hi
I think it means all power fets on heatsink, which is logical |
|
|
| wg_ski |
I build mostly big PA amps. Because I build exclusively from stuff obtainable on the surplus market, there isn't always the luxury of one big perfectly sized and machined heat sink for everything to mount to. MOst of the time, power transistor banks get split up and the NPN and PNP banks may see slightly different thermal environment. It looks like this could happen here as well. I've battled thermal runaway that sometimes takes DAYS of idling to set in and it's due to these slight temperature variations that are responsible.
To minimize the possibility here, mount D1 as close as possible to TR13, mount D2 next to the hottest of TR16-19, Mount D3 next to the hottest of TR20-23, and mount TR10 next to TR15. The closer the better. AS for D1-D3, do not use diodes. Use diode-connected transistors (B shorted to C) in fully-insulated TO-126 cases. You can put them anywhere that way, and not worry about thermal contact or getting cases or leads shorted to heatsink. |
|
|
| ricsmuts |
AHH, thank you :D
Now that's the answer I was after.
I once worked for a guy who built large PA amps, and you guys know the score.
Many thanks, will do exactly as you say. |
|
|
| EWorkshop1708 |
| quote: | Originally posted by ricsmuts
WOW, you don't go easy on the heatsinks! I don't think you will have any heating problems there:D
If you ever consider posting that thermal track design, look me up please.... |
It's still in the works, but when it's finished, I'll be glad to post the design. :) I'm still working out all the resistance values and the final design of the schematics and also components layout.
IMO There's nothing better than really large heatsinks to ensure reliable temps and avoid runaway. Even if you have little or no thermal compensation, a big sink will stabilize temp after a while, and usually will not thermally runaway.
*I recommend for a good heatsink, is a really thick mounting base, so you don't get hotspots, and the sink warms as a whole.
*Also, I recommend large and dense fins to carry the heat away faster, and allow use of a fan if needed. If you can't get one with nice big fins, or any fins at all, substitute for that by having a much larger mounting base instead. |
|
|
| luka |
| quote: | *I recommend for a good heatsink, is a really thick mounting base, so you don't get hotspots, and the sink warms as a whole.
*Also, I recommend large and dense fins to carry the heat away faster, and allow use of a fan if needed. If you can't get one with nice big fins, or any fins at all, substitute for that by having a much larger mounting base instead. |
This great solution to the prolem that shouldn't exist in the first place :D |
|
|
| ricsmuts |
Big heatsinks are always a good idea, but Its more important that the circuitry is first stable enough for even a small heatsink.
This is particularly important to me as this is a car amp that will be pushed extremely hard under less than the best conditions.
I am facinated by those Thermaltraks and am keen to see how they perform in a good design. |
|
|
| ricsmuts |
WG_SKI
Your fix worked like a charm, thanks.
Amp is stable and I can turn up the Bias now.
If I switch on the amp cold and immediately run a 100Hz sine wave at high volume, the current draw peaks for about half a second , but recovers without damage. Thermal lag in the heatsinks I guess.
(need to get the componets real close together as I dont want to stress the output transistors)
As it's a car subwoofer amp, I'm trying to simulate all the things people do to their amps. |
|
|
| ricsmuts |
| Ok, now the thermal issues are sorted, anybody have any advice as to increasing the drive to the output transistors slightly as I added some extra aoutput transistors when I did the PCB. |
|
|
| wg_ski |
| quote: | Originally posted by ricsmuts
Ok, now the thermal issues are sorted, anybody have any advice as to increasing the drive to the output transistors slightly as I added some extra aoutput transistors when I did the PCB. |
Experiment with R20 and R25. I haven't done a full analysis of the quiescent currents in this exact circuit, but I have plenty of experience with ones just like it. I usually bias TR9 and TR11 between 8 and 12mA and if you do that the triple darlington will drive a welding rod with enough outputs in parallel. They will need heat sinking. The jury is still out on whether R21 and R24 are necessary. Leach used to include them, but recent incarnations of the circuit have done away with them. Increasing those or eliminating them gives more available drive, but they are useful in tailoring the open-loop gain for stability. |
|
|
| ricsmuts |
You are a wealth of information:D
TR9 and 11 don't even get warm right now, even at full power.
Any suggestion as to minumum values for R20,25 |
|
|
| wg_ski |
| The minimum value I'd use for Re is at least 10 times the intrinsic emitter resistance (26/Ie in mA). For 8mA, it's about a 30 ohm minimum. But how much current is actually there with any given resistor is dependent on what's on the base side. Just reduce Re until you get between 8 and 12mA - more current isn't needed. Measure the drop and calculate what you need to change it to - if you need to at all. If your rails are only at +/-40V, they'll only get a little warm, but heat sinking will let them settle on final bias faster. At my typical +/-100 volt or higher rails they get toasty. They're supposed to run class A and run the same temp idling or at full blast. |
|
|
|
