Advice on setting bias... |
Post Reply | Page <1234> |
Author | |
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
Looks pretty neat, get yourself some RGB LED tape and have some mood stacks fading in and out of different colours slowly!
As to amp stuck on positive rail it would probably suggest one or more of the output devices attached to said rail has gone short circuit and is pulling the output up to there. Or the positive rail output devices are being pushed hard on for some reason. Standard repair procedure. Test output devices, remove from circit if necessary. Fault find and test PSU, then verify correct small signal circuitry operation through LTP and VAS to ensure correct operation. Finally replace output devices with working parts, bias to minimum, and carefully power up. A variac or light bulb limiter helps here. |
|
monkeypuzzle
Young Croc Joined: 25 January 2005 Status: Offline Points: 924 |
Post Options
Thanks(0)
|
I did originally think of doing that and putting a dmx controller on it and running it from my desk but in the end felt that the warm amber glow suits the age of the amps. What you suggest is pretty well what I've been doing but maybe not quite in the methodical way of a seasoned professional or with the ability to see cause and effect very clearly. I was unsure about running it up without the output devices on but I'll do that. In respect to following the signal, how does the signal from the feedback effect fault finding? I know that on this board here the very first section is correct as I've checked on a working unit a while back and have about +0.6v at the top of R7. On this board I can only get that when I pull one leg out of R11. With it in place I read about +27v at the top of R7 which I assume is the effect of the feedback loop skewing everything from that point onwards which in my simple head makes chasing it seem rather cyclical... When I get my tools back here I'll pop out the TO3s and work my way back along the signal path in a proper methodical manner. I've got a variac here but haven't properly used it. Is it a case of power at a low voltage and check that everything that should read 0v is and slowly increase? Oh and here's the question that shows my ignorance, what do you mean by LTP and VAS? My lampy head knows LTP as something and the speaker builder in me knows VAS as something, neither have anything to do with amplifiers.... I guess you're referring to the separate stages of the circuit? Cheers.
|
|
blah blah blah blah blah......
|
|
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
I'd get a micro such as an arduino or PIC and program it to run the LEDs myself, have a manual or auto mode where you set RGB channels and fade via potentiometers, and auto to do what you like. Neat thing about developing that sort of circuit yourself is that not only do you learn, but you then have the code to re-use and could for example make a neat RGB LED infinity table or similar. Anyways! You won't damage the amplifier with the output devices and feedback loop out, and then you can probe the small signal circuitry ideally with a scope for AC signals, but you can see DC bias points with the amp input grounded and that's a good start too. Your simulator will give you a very good ideaof what they should be. Once the feedback signal is connected, you have closed the previously open loop gain of the amplifier which overall will reduce it. Going off thelast schematic that Matrix posted your feedback tap is R11 C3 and R10. The feedback fraction is (R10+C3)/(R11+R10+C3), it will be greater at lower frequencies due to the roll off effect of the capacitor. You will want to ground the base of Q1B when the feedback tap is diconnected). (In my opinion) A well designed functional amp with matched transistors should give an output fairly close to 0V even without feedback connected and all transistors in, but the gain will be much higher so if you do test with a signal you'll have to keep the amplitude down a lot to avoid clipping at parts of the circuit. This really depends on the circuit in question. Connecting the feedback will help bring output offset even closer to 0V. If you dont know what the DC voltage bias points should be, use a working amp as a reference or simulate the circuit to get an idea. If something measures differently as you work through and build up the circuit that gives you a clue as to where to start digging. Slowly build it back up checking DC bias points as you go and then checking with injected AC signal. If I was unsure about the state of it being old, I'd disconnect the amp boards and check the PSU out first. Then I'd power the boards of a pair of lab DC supplies with current limiting ideally. I'd expect the amp boards would be working at +-20V DC or even lower. Then I'd reconnect PSU and run up from a variac or with light bulb limiter. LTP is a Long Tailed Pair, i.e. Q1A/Q1B. This amplifies the difference between the signal applied at the bases of each transistor,w hich is why the feedback tap comes back to Q1b's base. Q2/Q3 are an adjustable current source load for the long tailed pair, which boosts its voltage gain and has some other advantages over a plain resistor load. Output of the LTP is tapped by R6, which with no signal generates enough at standby voltage to bias Q6 on. Q6 is the VAS (Voltage Amplfier Stage) and is providing most of the voltage gain in the amplifier. This is the point that will probably clip first without feedback. The VAS is loaded by another constant current source of Q4/Q5, which again increases voltage gain and improves performance over a straight resistive load. The constant current from Q4/Q5 generates a voltage over D1/D2 which is going to be more or less constant regardless of current here, and then additional voltage is dropped over R14 + R15/R17. It is the voltage over these resistors and both diodes that biases the next pair of transistors on (Q9/Q10) so you would expect to see ~1.3-1.4V between collectros of Q6 and Q5 when all that is working at DC. Then the 3 pairs of output transistors provide voltage and current gain sufficent to drive the final big boy pair which are basically giving near zero voltage gain but a lot of current gain to the low impedance speaker load. The "real" bias of the amplifier is what current is flowing here at idle, which can be calculated by measuring the voltage over Vmeasured/(R33+R34 (0.3)) The power disipated in each transistor will be the bias current multiplied by its voltage drop from emitter to collector. This is where a scope helps, you want to tune R14 such that the AC output waveform measured at amplifier output loses its crossover distortion. You can go higher than that point but it means more power disipated in the output transistors at idle. Real bias is usually anywhere from 10-100mA depending on circuit and pesonal preference. At 100mA you will be dissipating 5W per output transistor, without a load even being applied. This is actually quite a lot of heat to sink! In a gig situation and a hot room it might go badly. For PA duty I would bias as low as realistically possible, for living room HiFi go nuts and experiment. But maybe watch the output heatsink temperature. You'll have to set the bias a few times over the course of an hour because temperature does affect it. Come back and adjust as necessary once everything warms to stability. If you find all this interesting, I would highly recommend a copy of : https://www.amazon.co.uk/Designing-Audio-Power-Amplifiers-Cordell/dp/007164024X Get a cheap second hand or ebook copy, the new price is a bit sharp. It is quite an approachable textbook though, you don't need a lot of maths to get sense out of it, but it won;t hurt either Right better go do some work Edited by odc04r - 06 August 2020 at 11:33am |
|
monkeypuzzle
Young Croc Joined: 25 January 2005 Status: Offline Points: 924 |
Post Options
Thanks(0)
|
Thanks you so much, thats really useful and very well explained. I can certainly work from that and it fills in the gaps in my knowledge perfectly. I managed to bring home a working one along with all my tools and have started to get some voltage readings across the working board.
Its going to be slow as I'm keeping the kids entertained, building a bar out of what ever I can find for my father in laws garage as a birthday surprise and trying to organise my wife's 40th! But... I will perceiver and post up my findings over the next week or so. In regards to the scope and checking for visible crossover distortion, I did try this when setting the bias originally with the scope watching the output and feeding in a sine wave. Even with the bias right down to zero I couldn't see anything change. This was trying the scope on 0.5v, 1v and 2v and feeding in a very low level signal. My scope is an ancient Telequipment D51 and is certainly not calibrated so I don't take any readings seriously but it should at least show the distorted part of the waveform around zero v with no bias shouldn't it?
Edited by monkeypuzzle - 07 August 2020 at 10:58am |
|
blah blah blah blah blah......
|
|
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
I have had projects on the go for multiple years, slow is the word. What's the rush. Probably what you are seeing is that the bias across the 2 diodes is sufficient to the bias the output devices on their own. You will get a minimum voltage across the diodes in forward conduction, and then it will change very little as you push the current up due to the diode forward characteristic. You could put in smaller forward drop dides such as Schottkeys instead, or remove one diode and size the resistor vs DC current to give you an adjustable voltage that will take you in and out of the bias region. The nice thing about diodes is that they are a semiconductor junction similar to a bipolar transistor, which is pretty much 2 diodes back to back. This means they can be advantages with regard to thermal tracking if implemented right, i.e. as output transistors get got the bias may also track to some extent, positively or negatively. But usually for this the bias diodes need to be in good thermal contact with something like the output heatsink, so I wouldn't worry about it here too much.
|
|
monkeypuzzle
Young Croc Joined: 25 January 2005 Status: Offline Points: 924 |
Post Options
Thanks(0)
|
Okay so I had a few moments to have a dabble...
I pulled out R11 and grounded the base of Q1B, popped out R161, D4 and Q10 to completely separate the front half from the second half and I think I might be getting somewhere.... Adjusting R9 gives between 0.3 and 0.5v across R6. I guess this mean that Q6 isn't biased correctly? I'm getting the +ve rail voltage on the collector of it. I've popped it out and it meters out correctly at 0.52 and 0.54 from the base. Cheers.
|
|
blah blah blah blah blah......
|
|
monkeypuzzle
Young Croc Joined: 25 January 2005 Status: Offline Points: 924 |
Post Options
Thanks(0)
|
Okay! One fixed.. Both of the bias diodes were shot... replaced all diodes on that board as a precaution. I've noted which one and I'll swap all the diodes out in its partner board when its not nearly midnight.
Going to have a quick look at the other amp I have sitting here and then game over for the day! Thank you very much odc04r and M4trix for your interest and advice, its really appreciated and has been what I needed to follow this through with a more thought out and professional workflow. |
|
blah blah blah blah blah......
|
|
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
Yeah that sounds about right, depends on the characteristics of a 2n5415 as to its turn on voltage. It'll be a medium power driver I expect so may turn on a little lower than 0.6V
Q6 only needs to steal enough base current at DC quiescent operation from R6 to get enough current through D1 and D2 as a first test. Depending on the current Q4 and Q5 are pulling then the base current of Q6 will be that figure divided by its gain (beta or Hfe in datasheets). If you measure ~1.3V over D1 and D2 its all probably working ok. If you want more voltage over R6, then the Q2/Q3 current source can be altered or R5 also looks like it will be a part of it. R5 is going to see current of 20/R5 assuming negligible in ZD1 (20V is set from ground by the Zener ZD1) or 20/1000 = 20mA But then Q3 collector-emitter current is set by the combination of R8//R9 in parallel with the ~0.6V over Q2's base when on. If R9 was about 50% on then you'd be looking at about 250 Ohms in parallel, or 0.6/250 = 24mA. I suspect the ~4mA difference is probably lost via the Zener diode. Odd way to make the current source really, I can only assume that to drop all the voltage of the -tive rail over the BC182BPs would stress them to much, so the Zener diode/resistor parallel combo pushes them up a bit for an easy life. So there's a few more things to measure anyway. |
|
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
Cool, they may have failed because their current source went doodah so check Q4/Q5/R16/R12 also and verify current by Vbe (Q4)/R16
|
|
monkeypuzzle
Young Croc Joined: 25 January 2005 Status: Offline Points: 924 |
Post Options
Thanks(0)
|
Ace, will look over the other parts of the circuit and read through your remarks properly later this week. On thing to note is that the original circuit has a 20v Zener, all mine have 22v. What difference would this make overall?
Today however I've a busy day pushing a flightcase from Manchester Academy to St Peters square for #wemakeevents then off to the unit to pull out all my Palcos and a few projectors for the http://www.lightitinred.co.uk/ camapign. Standard weather for an outdoor gig....
Edited by monkeypuzzle - 11 August 2020 at 9:33am |
|
blah blah blah blah blah......
|
|
odc04r
Old Croc Joined: 12 July 2006 Location: Sarfampton Status: Offline Points: 5482 |
Post Options
Thanks(0)
|
Buggery do really. A very slight change in current across the tail resistor in parallel with the Zener. Long as it is 20mA ish, then it's fine. Oh yes, tomorrow could be good for an outdoor gig. I have a week of holiday starting then too. Pretty much semi-constant rain for an entire 7 days. Still, it's not work.
|
|
M4trix
Registered User Joined: 23 June 2009 Location: Croatia Status: Offline Points: 445 |
Post Options
Thanks(0)
|
@monkeypuzzle, looks like, with a huge help from odc04r, you're sorting things out ! Good.
|
|
Post Reply | Page <1234> |
Tweet |
Forum Jump | Forum Permissions You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |