Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
So, something must be escaping my attention here, somewhere.
I replaced R14 (with a higher wattage resistor actually, 1/4W instead of 1/8 I guess). I reconnected R11, R12, lifted R15, R16.
I soldered Q7, Q8 and Q9 back in. So as you said, only R20, R15 and R16 are lifted. The amp switches on fine, but... ZD1 = 0V, Q7 E-C = 0V... (At least it didn't short a fuse.) I'm confused.
I bought a spare Q5 and Q6 in case, as well as some diodes and more resistors.
Missed something obvious. R16 and R15 have to be in to get voltage across ZD1.
Let's add a stage and see what happens. Remove Q10, Q11, R20 is out. Jumper can stay in for Q19.
Put a meter across R17 (220 ohm) before you power up.
If you don't have < 4V; turn the amp off immediately.
See if any transistors are hot.
You should have voltage across ZD1.
Q19 - Emitter to COLLECTOR; remember I made a brief mistake which I immediately corrected, The base current modulates the E-C junction or E-C turns on with enough base current. The purpose of shorting E-C is so that you don;t have to apply audio,
When we shorted C4, the purpose was was to make the AMP amplify DC so it could be tested with a battery Shorting E-B of Q19 COULD DAMAGE Q19!!!
Q8/Q9 and Q9/Q10 is basically a darlington pair. This configuration multiples the current gains of the devices
So when you look at the circuit:
Zd1 and R6 and Q3 forms a current source.
Q1, Q2 and Q3 is a differential amplifier which requires a constant current source
The differential amplifier also phase splits.
Q4 is basically an inverter
R11 and the mess through R12 is basically a bias regulator. Because Q7 is mounted on the heat sink of the output transistors it does temperature compensation of the bias.
The bias regulator's purpose is to maintain a fixed voltage between the bases of Q10 and Q11 so that the AMP operates in class AB mode.
The stuff past that starting with R20 is DC protection except for C11, R22 which is a Zobel network.
No variac, scope, ammeter, signal generator makes things tough.
See what happens when:
the EMITTER and COLLECTOR od Q19 is shorted,
R20 out,
Q10 and Q11 out.
Voltage across R17 when powering up. I think it should be about 1.2 Volts, Lower is OK.
Wow. I have lots of things to google and lookup. It may be a "simple" circuit, but it's pretty hard for me to read, as I'm not yet familiar with how everything functions. I have quite a few messy print-outs now, with coloured markings everywhere!
So, E-C Q19 shorted, R20 lifted, Q10 and Q11 out:
R17 = 0V
ZD1 = 5.5V
Does it make sense for R17 to be at zero? Situated between Q8 and Q9...
OK, this is getting very interesting. First thing I noticed, using exactly identical conditions as previously (I double check my readings now!), was that whilst switching the amp on, R17 voltage jolted to 1V approx, and rapidly dropped to 0V.
Now, I short C4... When I switch the amp on, R17 voltage stays at 0 for a few seconds, then very progressively increases, until it reaches about .92V (took a good minute or two), and remains stable then. The lot has been on for 4-5 min now, no sign of smoke. Q1,2,4 area is mildly warm, hardly anything alarming though (I don't think!).
EDIT : actually, I think the jolt might be a few volts before returning to 0 (first case scenario). I'm still waiting for my multimeter upgrade, this one doesn't have autorange, it's just an on screen estimate...
EDIT2 : the jolt, with C4 NOT shorted, is probably symptomatic of my original problem, the "thumping" when the sub is started (and when I switch channel on the source; currently those tests are done with no source of course.); interestingly, it only happens when the sub is switched on from cold. If I switch it off, leave it 20-30 seconds (for the red LED to die off), and switch it on, I get no major "thump" or voltage jump. The longer I leave it off for, the bigger the jolt upon startup. (Although this is probably something to discuss once the circuit no longer shorts itself!)
OK, before we go any further, let's switch gears a bit.
Look at the voltage across R2||C2 as you power up and as you power down
C1, C10 and C17 are actually the wrong type cap. C2 is too.
I'd use a 100 uf NP cap for C2 and C4. e.g **broken link removed**
I'd use something like **broken link removed** for C1 and C17
These are the sorts of coupling caps that manufacturer's skimp on. They should be non-polarized and non-polarized are expensive.
I think you will have to order them.
Assuming Q10 and Q11 are good and in right and R18 and R19 are good, I think the amp may power up with everything connected and not smoke.
Before you do, Leave the jumper across C4 and have R20 lifted; put Q10 and Q11 back in.
Place a voltmeter across R18 or R19.
The voltage across these resistors will be in the millivolt range. 5-20 mV. If much higher than that turn things off.
If things seem to work, then put things to rest for a while until you get non-polarized replacements for C2, C4 (100uf) and C1, C10 and C17 (4.7 uf).
Replace these. The last things you need to look at will be:
The DC voltage across the speaker terminals. <75 mV is a good number. 5-10 mV is a great number.
The voltage across R18 or R19 should not run away as the amp heats up. This is very dependent on the thermal mounting of Q8, Q9, Q10, Q11 and Q7. Q7 actually measure the temperature of the heatsink.
With Q4 shorted, Q19 E-C shorted, R20 lifted, and Q10-Q11 still out, I get the following measurements:
AC switched on : R2 || C2 = 0.03V. The voltage is pretty much instant, and stable.
AC switched off (after power up) : for a few seconds the voltage ever slighly increases, 0.032, 0.034 etc; then, after those few seconds, it shoots up for another 2-3 seconds, until it reaches approx. 0.36V at least. It then proceeds to reduce as quickly as it shot up, down to 0.
C1, C10 and C17 are actually the wrong type cap. C2 is too.
I'd use a 100 uf NP cap for C2 and C4. e.g 100uF 100V Non-Polarized Capacitor
I'd use something like Mallory Axial Polyester Film Capacitor (Mylar) | 100V/4.7UF/A/MY (100V4.7UFAMY) | Mallory for C1 and C17
These are the sorts of coupling caps that manufacturer's skimp on. They should be non-polarized and non-polarized are expensive.
This is really good to know... Not that surprising they skimp on parts, but well, any improvements are also welcome additions. I'll order those. Do you think this might partly explain the thumping problem? Or is this just an additional improvement, while we are at it?
OK, I turned it on, and the voltage carried on increasing... It quickly got to 60mV (R19 reading) and carried on increasing, so I immediately turned everything off. Q10 and Q11 got warm within a few seconds, but hopefully I didn't "toast" them this time, no magic smoke, but clearly something is up. Could R18 or R19 be damaged? The in-circuit resistance seems to be around 0.7-0.8 Ohm each.
I tested Q10 and Q11 out of circuit before putting them back in, and they seemed to be fine, readings around 0.560V where needed (in diode mode), and open when needed. They don't seem to be open in anyway... That leaves us with R18/R19?
The DC voltage across the speaker terminals. <75 mV is a good number. 5-10 mV is a great number.
The voltage across R18 or R19 should not run away as the amp heats up. This is very dependent on the thermal mounting of Q8, Q9, Q10, Q11 and Q7. Q7 actually measure the temperature of the heatsink.
So, Q7 measures the temperature of the heatsink, to adapt the current to Q10-Q11 temperature?
"Thermal mounting of Q8, Q9, Q10, Q11 and Q7" - those first four transistors seem to be critical, and Q10-Q11 specfically, quickly overheat. Not sure about Q8-Q9, I didn't want to leave it on any longer...
In the meantime I will order those better quality caps. Also, a good bit of news (could come in handy), I have a new multimeter underhand. It's a Velleman DVM68, it has much better functionality, readings, autorange, data-hold etc; so this might come in handy (as well as the fact I can potentially use 2 voltmeters at the same time now).
The DVM is pretty cool. I have an early Fluke 77 and one other. I have a FET Analog DVM that I built from an EICO kit (35+ years ago). I have some various HP, and Keithley meters such as the Keithley 195.
So, Q7 measures the temperature of the heatsink, to adapt the current to Q10-Q11 temperature?
This bias is part of the definition of class AB. There needs to be some conduction of Q10 or Q11 at all times. This eliminates what's called cross-over distortion or a gap when the signal crosses zero. If Vbe of Q10,Q11 is not temperature compensated, then thermal runaway could occur. There are a few special transistors that actually have a diode embedded in the package.
Check C2 in circuit in resistance mode. The caps could have something to do with thumping.
This is the FIRST amp that I have seen that does not have a potentiometer to adjust the bias, so I'm thinking that this amp uses the presence of audio to effectively turn off the amp.
So, I'm suggesting to put the amp completely back together except don't connect the speaker leads.
Monitor the voltage across R18 or R19. It should be zero with no signal because Q19 will be off and C4 will be off.
Play some music and keep an eye on the DC value on R18 or R19. Since you have second DVM now, look at the DC voltage at the speaker terminal.
As for the transistors, an A935 is a 2SA935. 2SA, 2SB, 2SC, 2SD ..., Japanease diodes typically have a prefix of 1S
You might want to consider, Replacing R14 with a 3.3K resistor and putting a small 500 ohm 10 turn pot where the wiper feeds the base of Q7. Something to think about as well as matching Q1 and Q2.
Been busy today with both inside, outside and errands.
This bias is part of the definition of class AB. There needs to be some conduction of Q10 or Q11 at all times. This eliminates what's called cross-over distortion or a gap when the signal crosses zero. If Vbe of Q10,Q11 is not temperature compensated, then thermal runaway could occur. There are a few special transistors that actually have a diode embedded in the package.
Does this explain why the R19 voltage kept on increasing beyond 60mV? I need to have the heatsink attached? I still found it quite surprising, it seemed like a very rapid increase in voltage in temperature, I fear burning something again. Or could the overheating be due to the capacitors?
C2 (in circuit) = 21.6 kOhm
C6 (in circuit) = 2.65 MOhm??
This is the FIRST amp that I have seen that does not have a potentiometer to adjust the bias, so I'm thinking that this amp uses the presence of audio to effectively turn off the amp.
The is the temperature / voltage coefficient? Is this why you are suggesting to add a 500 ohm 10 turn pot at Q7? This could be useful, at least for controlling voltages during testing and making sure I don't burn anything again! I got a spare 1/4W 3.3k resistor for R14 - I assume this would only have to be changed once a pot is installed, or would that help now? Of course the other PCB has a gain and crossover pot, but I assume those are totally independent.
As for the transistors, an A935 is a 2SA935. 2SA, 2SB, 2SC, 2SD ..., Japanease diodes typically have a prefix of 1S
Here is some stuff to look at: **broken link removed**
Everything back in, including R20. Old caps, R14 unchanged.
In stand by mode (red led), we have :
R18 = 0V
R19 = 0V
In forced on mode (green LED):
R18 = instantly reached 20mV, and kept on increasing to 100mV, by which time I killed the power...
R19 = same as above.
I didn't get around to testing with music, Q10-Q11 got very hot, very quickly. This is with the PCB on the desk, not mounted back onto the aluminium plate with the heatsink, but still, this is getting too warm much too quickly, which wasn't the case when I first started trying to identify the cause of the "thumps". Why are Q10-11 heating up so quickly, with R18-R19 voltages ever increasing?
C1, C10 and C17 are actually the wrong type cap. C2 is too.
I'd use a 100 uf NP cap for C2 and C4. e.g 100uF 100V Non-Polarized Capacitor
I'd use something like Mallory Axial Polyester Film Capacitor (Mylar) | 100V/4.7UF/A/MY (100V4.7UFAMY) | Mallory for C1 and C17
I got some new caps, hopefully those are the right ones; if not let me know, I can return them to the shop, I haven't soldered them in yet. I have 2 Philips 4.7µF 100V - MKT caps, and 2 "BP" SOLI 100µF 100V caps, usually used for speakers etc. They are non-polarised apparently. I've attached a picture below; if those are not the right sort, I'll return them and get the precise ones online (they had up to 2.9µF at 100V Mylar only in the shop, and no thin polyester film ones).
Basically, the junction temp of Silicon must not be exceeded. Usually that's about 150 deg C. There is usually a spec for Junction to case, a spec for the mica washer and a spec for the heatsink (color and and area matter) and an ambient temperature. From the expected power disipation, you can size a heat sink. Usually you'll select a heatsink based on what you need and manufacturer's tables.
If you have ever taken apart a large metal transistor, you;d see that the die isn't very big.
The amount of thermal grease that you use is very small, basically to fill imperfections in the metal. The type of thermal washer that's used also matters, so it doesn't take much for a transistor to "run away" thermally. In general, you need termally conductive and electrically non-conductive.
Shorting Q7 (E to C) should prevent the transistors from heating up although it doesn't fix the problem. Mounting them back on the heatsink MIGHT fix the problem. 0V across E-C of E-C of Q7 disables the bias regulator. I'm not sure what it should be in this case. I THINK the voltage across Q7 (E-C) should be about 2.4 V when working properly and anything less is nearly off. It's either around 1.2 or 2.4.
The switch has three positions:
OFF - Red LED
Auto - Red LED with no signal, goes green when it detects a signal, film, song etc
ON - green LED - forced on (signal or no signal)
Third: The caps are a better choice than what you have.
Yep... Which worries me. Before I went ahead shorting things whilst taking voltage measurements, I had the PCB out, detached from the heatsink, and I could leave the amp on fine, those wouldn't overheat. Now they get very warm, very quickly... Could it be something wrong with Q7? I replaced it, maybe I shouldn't have? I have the old one if need.
I can try remounting everything with the heatsinks etc. but I'm not too optimistic that it will solve the problem, I think the heat is really excessive unfortunately. I'm wondering why Q7 isn't serving its purpose and regulating the voltage, when it used to ! (On a side note, I have some thermal paste, which is eletrically non conductive - I have a fair bit from CPUs!)
I'll do some voltage measurements and try shorting Q7 is things get overkill, but as you say, I don't think that would be a long term solution... We'd have no thermal adjustment then. At least it seems we are getting closer to identifying the magic smoke issue !
Hfe between 200 and 700 is too wide for this application. So try for 0.1 mA of base current and 1000 x would be 1 mA of collector current. Can't go above 20 ma for base current and no more than 100 mA of collector current.
The test circuit should look like; Rb in series with the base, connected to the 1.5V battery and the emitter.
Take a 9V battery and put a resistor, Rc in series with the battery and the collector. The - of the 9V battery should go to the emitter as well. Put voltmeters across Rb and Rc.
Using your battery (1.5 V); the base resistor should be about (1.5-0.6)/0.1e-3 amps.
Multiply this by 1000 (700 with a fudge factor, easy # to multiply)
Take the recalculated current, Ib; (1.5-0.6)/Rb and multiply by 1000.
Size Rc*(ib*1000) to be 5-8 volts. Not critical.
Measure Rb and Rc.
Measure Vb (voltage across Rb), Rc (Voltage across Rc) and make a table.
Transistor n | Rb | Vb | Ib | Rc | Vc | Ic | Ic/Ib
Rb and Rc are fixed
Select the transistors for Q1 and Q2 which have the closest Ic/Ib ratio. Do write down the original Ib/Ic for Q1 & Q2. Now that you've done all of this work keep the transistors labeled by Hfe.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
