@Tom_S said:
I’m getting pretty close to firing this one up. I’m using Tony’s upgraded transistor substitutions - had to use a Ksa992 for the 2sa773 he used in one position. Pinout is different and had to do the leg twist, so we’ll see how that works out.
All that’s left are the TO-3s.
Hey Tom, I was looking at your board and noticed that you have yet to install jumpers J1 and J2. They are really small and easy to miss. Just thought I'd mention it, just in case.
Interesting how the pcb designer layed out all the parts in nice neat rows. I have never seen that before. Usually you try to keep traces as short as possible and not have multiple traces running along aside each other. Looks like this is a one sided board.
Good catch! I hadn't even noticed that I missed those last night. I got the outputs installed this morning. I really hate messy thermal compound, but silpads are just too expensive for an experiment like this. Just need to locate my transformer and we'll see if I pass the smoke test.
@PWRRYD said:
Interesting how the pcb designer layed out all the parts in nice neat rows. I have never seen that before. Usually you try to keep traces as short as possible and not have multiple traces running along aside each other. Looks like this is a one sided board.
Yes, it is one sided. And many of the traces are very thin and run long distances along side each other. Does not seem to have a hum problem, though.
@PWRRYD said:
What is needed for a transformer (+/- V?, VA?).
I fixed the crossover notch problem and am power testing the amplifier right now. My bench supply can do 5 amps up to +/- 30vdc, which is more than adequate. But I don't want to test it at much beyond +/- 18 to 20vdc or so. The transistors are only rated for about 30vdc. Power supply caps are rated 35vdc.
One of the big problems appears to be the small, little heat sinks used on the TO-3 transistors. With it powered at +/- 18vdc, I fed a 0.1Vrms (100mv) 1kHz sine wave into the input. This produced 12 watts rms into an 8 ohm dummy load. Total current draw for 1 channel was 850ma, so 2 channels would need 1.7amps to produce 12 watts rms. But after about 2 to 3 minutes I had to power it down. The output transistors were going into thermal runaway, one transistor measured 140F and the other was about 129F and climbing.
Yeah I've been doing some light reading on the subject. One issue of BJTs is they tend to draw more current the hotter they get, hense the runaway. Thay is why Mosfets became popular. They tend to be more self regulaing from temperature.
In the video, Tony moved one of the bias tracking diodes over to one of the heatsinks and bonded it in place with heatshrink tubing. In this way, the biasing diode will track with the heatsink temp and prevent thermal runaway. But the stock amp places these diodes quite a distance from the heatsinks, so there is no thermal tracking.
Fixed my Left channel issue. I threw the stock transistors into that channel first, but found that wasn't the issue. But did find it while I was soldering those in. I had pads on R5 & C6 that were not making connections. My own fault, since I missed the fact that I had my soldering station cranked up to ~800 degrees from a previous project. Must have cooked the traces underneath the solder before I realized my iron was so hot.
I now have decent looking sign waves. DC offset is around 50mv. No crossover distortion, but I do have lots of hum using a 180VA 18-0-18v toroidal transformer. This thing has way too much gain. I let it run for about a minute, feeding it 400Hz/18mv from the signal generator, which put the outputs right at 1 watt. The transistors and heat sinks were already getting pretty warm. I got up to 160 degrees on one of them and shut it down.
Ya, that is one big donut, you certainly won't be starving the amp for current with that thing! I've been soldering at 700F and have had two cracked and lifted solder pads so far. One of the pads on the small 3 terminal audio signal connector at the input and one of the pads on the speaker output connector. And when you put wires on these connectors, you have to be very careful not to bend the wires too much, because this can pull the pads loose. I had to install a jumper wire from my broken speaker output connector directly to one of the output transistor posts. Not a very high quality circuit board, that is for sure.
I was thinking that maybe I am not experiencing hum because I am using my bench DC power supply going through the four rectifier diodes. The four 1000uF caps therefore do not have to do any filtering to reduce the hum. When I switch over to a regular AC transformer, I'll probably start getting some hum.
Yes - we might need to add some extra filtering and probably fine some beefier diodes while we're at it. I'm still not sure what I'm going to do about a case. The there's the heatsink. Not sure I want to try to mill a spot for a TO-3 in the ones I have. The way this one is going, it makes building Gainclone chip amps in the early 2000s seem super simple.
Yup. Looks like we might have to drop back 15 yards and punt on this project. Things are not looking good. In its stock form, this kit is completely unusable. And the nature of the mods needed to make it workable will probably run the cost up way too high. Probably not worth spending the money on a cheap $15 kit. I will probably continue to work on my units to see what can be done at a reasonable cost. But the heatsinks are simply too small. In its stock form this kit will not even handle a continuous 1.1w rms output per channel. I just did a power test, both channels driven at 1.1watts rms with a 1kHz sine wave, and I had to shut it down after 17 minutes because the outer two power transistors were approaching 150F !! Maybe if I replaced and moved the bias tracking diodes onto the heatsinks like Tony did in the video, this would solve this runaway heat problem. I do not know.
EDIT: But this project has been a good learning experience, even though it looks like it might turn out to be a complete failure.
Yes, the forward voltage drop across a diode changes with temp. You will see these type of compensation diodes mounted on the heatsinks of many power amplifiers using bi-polar output transistors. As the temp on the heatsink increases, the change in forward voltage drop across the diode automatically decreases the bias current in the output stage. If engineered properly, this prevents thermal runaway that could fry the transistors.
I watched Tony's older videos and learned 18/0/-18v should be the supply voltage, not the transformer secondary. A 15-0-15v secondary would be about right, but I don't think I have any transformers that will work in the parts bins. I agree, I think I'm going to toss in the towel on this project for now.
I agree. Sometimes when you take a break from a project, a light bulb goes off and you solve a problem or find a mistake. One thing that confuses me is that in the 6 year old videos, Tony is testing the same board that we are using with it set at the proper +/- 18vdc, but he did not have the runaway overheating problem. Even when testing it for an extended time period with both channels outputing a continuous 7 watts rms, the heatsinks only got moderately warm! And he did not have tracking diodes attached to the heatsinks for this test. He only had the 100pF caps and 470 ohm resistor mods installed. So I must be doing something wrong. Either that, or I have a set of 4 defective output transistors. Newark is running a sale on Multicomp Pro 2N-3055 transistors for a little over a buck each. I'm thinking about swapping in a set of these to see what happens. Maybe they can take a higher current without overheating, who knows.
For transformers, I have it narrowed down to this one for about $25 from Mouser. This is the one that Tony used at the beginning of the 6 yr old video: 115V 24vct 3.3a 80VA 3H x 2.5W x 2.31L. When he connected it, it provided the necessary +/- 18vdc voltage after the rectifiers when the amp was idling. https://www.mouser.com/ProductDetail/Triad-Magnetics/VPS24-3300?qs=VZvlzY8g9V8TItDCoz/WeA==
I wondered about those output transistors. Since we are both having issues with them running hot, you have to question whether they might be lower spec parts that got relabeled. I'm sure there's a more modern, bullet-proof replacement at Mouser, but I'm not sure I want to take the chance and sink more money in it.
Bought a set of 2N3055G's from Mouser to replace the outputs (Onsemi types). I'm also in the prcess of replacing the heatsinks. The new heatsink will have approx 6 times the thermal mass of the 4 smaller ones. It is a 1/8" thick by 2" aluminum angle, 12 inches long. The plan will be to drill and bolt this angle to an aluminum chassis for even greater heat dissipation. Since the base of these transistors is the collector, I need to mount the new transistors with mica insulators and nylon bushings to keep the collectors from shorting out. But that is no big deal, just a few extra parts.
Not to diss on Bill's (or anyone's) desire to experiment, learn, play, and have fun with this type of cheap "kit" but IMHO building a LM3886 gainclone DIY amp makes more sense. There is everything from simple 7 parts/ch designs to the moon. The single biggest cost is the power supply transformer but that really holds true for any amplifier design. Antek has some reasonably priced toriods for this. And Steve at Apexjr.com use to have some as well, and some great power supply caps too.
@PWRRYD said:
Not to diss on Bill's (or anyone's) desire to experiment, learn, play, and have fun with this type of cheap "kit" but IMHO building a LM3886 gainclone DIY amp makes more sense. There is everything from simple 7 parts/ch designs to the moon. The single biggest cost is the power supply transformer but that really holds true for any amplifier design. Antek has some reasonably priced toriods for this. And Steve at Apexjr.com use to have some as well, and some great power supply caps too.
For sure! I love these threads that Bill and Tom are doing.
Silver1omo,
That neurochrome amp is said to be stellar! He got every last bit of performance from the chips. I wouldn't hesitate to build that if the power output is what you need.
Comments
nasty
Hey Tom, I was looking at your board and noticed that you have yet to install jumpers J1 and J2. They are really small and easy to miss. Just thought I'd mention it, just in case.
Interesting how the pcb designer layed out all the parts in nice neat rows. I have never seen that before. Usually you try to keep traces as short as possible and not have multiple traces running along aside each other. Looks like this is a one sided board.
Good catch! I hadn't even noticed that I missed those last night. I got the outputs installed this morning. I really hate messy thermal compound, but silpads are just too expensive for an experiment like this. Just need to locate my transformer and we'll see if I pass the smoke test.
What is needed for a transformer (+/- V?, VA?).
Yes, it is one sided. And many of the traces are very thin and run long distances along side each other. Does not seem to have a hum problem, though.
I fixed the crossover notch problem and am power testing the amplifier right now. My bench supply can do 5 amps up to +/- 30vdc, which is more than adequate. But I don't want to test it at much beyond +/- 18 to 20vdc or so. The transistors are only rated for about 30vdc. Power supply caps are rated 35vdc.
One of the big problems appears to be the small, little heat sinks used on the TO-3 transistors. With it powered at +/- 18vdc, I fed a 0.1Vrms (100mv) 1kHz sine wave into the input. This produced 12 watts rms into an 8 ohm dummy load. Total current draw for 1 channel was 850ma, so 2 channels would need 1.7amps to produce 12 watts rms. But after about 2 to 3 minutes I had to power it down. The output transistors were going into thermal runaway, one transistor measured 140F and the other was about 129F and climbing.
Here are two transformers on my list of possibilities. The Triad one is 80VA, the Jameco one is 56VA:
https://www.mouser.com/ProductDetail/Triad-Magnetics/VPS28-2800?qs=XQhEZXYA1Z3IMS84kJPS2w==
https://www.jameco.com/z/PT-2866-R-Jameco-ReliaPro-28VCT-Power-Transformer-2-Amp-115-230VAC-Primary_149797.html
Yeah I've been doing some light reading on the subject. One issue of BJTs is they tend to draw more current the hotter they get, hense the runaway. Thay is why Mosfets became popular. They tend to be more self regulaing from temperature.
In the video, Tony moved one of the bias tracking diodes over to one of the heatsinks and bonded it in place with heatshrink tubing. In this way, the biasing diode will track with the heatsink temp and prevent thermal runaway. But the stock amp places these diodes quite a distance from the heatsinks, so there is no thermal tracking.
Right channel looks good on mine. Passes a clean sign wave at 12V DC measured at the diodes. The Left has an issue. I'll have to investigate.
Those transformers look pretty small. Maybe ok for 15 wpc.
Fixed my Left channel issue. I threw the stock transistors into that channel first, but found that wasn't the issue. But did find it while I was soldering those in. I had pads on R5 & C6 that were not making connections. My own fault, since I missed the fact that I had my soldering station cranked up to ~800 degrees from a previous project. Must have cooked the traces underneath the solder before I realized my iron was so hot.
I now have decent looking sign waves. DC offset is around 50mv. No crossover distortion, but I do have lots of hum using a 180VA 18-0-18v toroidal transformer. This thing has way too much gain. I let it run for about a minute, feeding it 400Hz/18mv from the signal generator, which put the outputs right at 1 watt. The transistors and heat sinks were already getting pretty warm. I got up to 160 degrees on one of them and shut it down.
More work ahead...
Ya, that is one big donut, you certainly won't be starving the amp for current with that thing! I've been soldering at 700F and have had two cracked and lifted solder pads so far. One of the pads on the small 3 terminal audio signal connector at the input and one of the pads on the speaker output connector. And when you put wires on these connectors, you have to be very careful not to bend the wires too much, because this can pull the pads loose. I had to install a jumper wire from my broken speaker output connector directly to one of the output transistor posts. Not a very high quality circuit board, that is for sure.
I was thinking that maybe I am not experiencing hum because I am using my bench DC power supply going through the four rectifier diodes. The four 1000uF caps therefore do not have to do any filtering to reduce the hum. When I switch over to a regular AC transformer, I'll probably start getting some hum.
Yes - we might need to add some extra filtering and probably fine some beefier diodes while we're at it. I'm still not sure what I'm going to do about a case. The there's the heatsink. Not sure I want to try to mill a spot for a TO-3 in the ones I have. The way this one is going, it makes building Gainclone chip amps in the early 2000s seem super simple.
Yup. Looks like we might have to drop back 15 yards and punt on this project. Things are not looking good. In its stock form, this kit is completely unusable. And the nature of the mods needed to make it workable will probably run the cost up way too high. Probably not worth spending the money on a cheap $15 kit. I will probably continue to work on my units to see what can be done at a reasonable cost. But the heatsinks are simply too small. In its stock form this kit will not even handle a continuous 1.1w rms output per channel. I just did a power test, both channels driven at 1.1watts rms with a 1kHz sine wave, and I had to shut it down after 17 minutes because the outer two power transistors were approaching 150F !! Maybe if I replaced and moved the bias tracking diodes onto the heatsinks like Tony did in the video, this would solve this runaway heat problem. I do not know.
EDIT: But this project has been a good learning experience, even though it looks like it might turn out to be a complete failure.
Definitely move the diodes to the heatsink or even on top of the transistors
For the lehman, do the diodes' characteristics change when heated? This is something I did not know....
InDIYana Event Website
Yes, the forward voltage drop across a diode changes with temp. You will see these type of compensation diodes mounted on the heatsinks of many power amplifiers using bi-polar output transistors. As the temp on the heatsink increases, the change in forward voltage drop across the diode automatically decreases the bias current in the output stage. If engineered properly, this prevents thermal runaway that could fry the transistors.
I watched Tony's older videos and learned 18/0/-18v should be the supply voltage, not the transformer secondary. A 15-0-15v secondary would be about right, but I don't think I have any transformers that will work in the parts bins. I agree, I think I'm going to toss in the towel on this project for now.
I agree. Sometimes when you take a break from a project, a light bulb goes off and you solve a problem or find a mistake. One thing that confuses me is that in the 6 year old videos, Tony is testing the same board that we are using with it set at the proper +/- 18vdc, but he did not have the runaway overheating problem. Even when testing it for an extended time period with both channels outputing a continuous 7 watts rms, the heatsinks only got moderately warm! And he did not have tracking diodes attached to the heatsinks for this test. He only had the 100pF caps and 470 ohm resistor mods installed. So I must be doing something wrong. Either that, or I have a set of 4 defective output transistors. Newark is running a sale on Multicomp Pro 2N-3055 transistors for a little over a buck each. I'm thinking about swapping in a set of these to see what happens. Maybe they can take a higher current without overheating, who knows.
https://www.newark.com/multicomp-pro/2n3055/transistor-npn-60v-15a-to-3-rohs/dp/81AH9584
For transformers, I have it narrowed down to this one for about $25 from Mouser. This is the one that Tony used at the beginning of the 6 yr old video: 115V 24vct 3.3a 80VA 3H x 2.5W x 2.31L. When he connected it, it provided the necessary +/- 18vdc voltage after the rectifiers when the amp was idling.
https://www.mouser.com/ProductDetail/Triad-Magnetics/VPS24-3300?qs=VZvlzY8g9V8TItDCoz/WeA==
I wondered about those output transistors. Since we are both having issues with them running hot, you have to question whether they might be lower spec parts that got relabeled. I'm sure there's a more modern, bullet-proof replacement at Mouser, but I'm not sure I want to take the chance and sink more money in it.
Bought a set of 2N3055G's from Mouser to replace the outputs (Onsemi types). I'm also in the prcess of replacing the heatsinks. The new heatsink will have approx 6 times the thermal mass of the 4 smaller ones. It is a 1/8" thick by 2" aluminum angle, 12 inches long. The plan will be to drill and bolt this angle to an aluminum chassis for even greater heat dissipation. Since the base of these transistors is the collector, I need to mount the new transistors with mica insulators and nylon bushings to keep the collectors from shorting out. But that is no big deal, just a few extra parts.
In for a penny ...
Not to diss on Bill's (or anyone's) desire to experiment, learn, play, and have fun with this type of cheap "kit" but IMHO building a LM3886 gainclone DIY amp makes more sense. There is everything from simple 7 parts/ch designs to the moon. The single biggest cost is the power supply transformer but that really holds true for any amplifier design. Antek has some reasonably priced toriods for this. And Steve at Apexjr.com use to have some as well, and some great power supply caps too.
I have been wanting to build this one:
https://neurochrome.com/collections/power-amplifiers/products/modulus-86-stereo-kit
He recommends this PSU:
https://connexelectronic.com/product/smps300reh/
I definately enjoy seeing folks messing with the fringe stuff though. Typically a niche I would find myself dabbling in.
For sure! I love these threads that Bill and Tom are doing.
Silver1omo,
That neurochrome amp is said to be stellar! He got every last bit of performance from the chips. I wouldn't hesitate to build that if the power output is what you need.