For a basic Gainclone, there used to be a ton of boards available. I may even have a set hanging around from Chipamps.com. I designed my own boards 2 decades ago, but I can't remember who I used - maybe PCBWay? I also used a surplus transformer from Apex Jr. It's a solid 55w/ch amp with no frills at all.
@PWRRYD said:
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.
Yeah, I'm getting close to concluding that the Neurochrome 686 is going to be my reference amp when I get my reference speakers done (hopefully by SDC 2024).
@Tom_S said:
For a basic Gainclone, there used to be a ton of boards available. I may even have a set hanging around from Chipamps.com. I designed my own boards 2 decades ago, but I can't remember who I used - maybe PCBWay? I also used a surplus transformer from Apex Jr. It's a solid 55w/ch amp with no frills at all.
That looks great Tom.
I bought pair of LM3886 chip amp kits from Aliexpress. Luckily mine showed up with real LM3886 chips, not the sometimes fake chips we heard stories about. They are bare bone circuit pcb's that don't have the Zobel or Theile circuits. Those are easily added. The basic pcb has a decent layout. I bought a 22-0-22 V Antek toroid transformer for them a few years ago and a bunch of PS caps from Apexjr. I should get off my rear and build a stereo amp with all that.
Thanks for the gainclone feedback. The main reason I started this thread, however, was because xraytonyb recently released a 10 part video training series based on this cheap OCL kit. So I thought that it would be a good way to increase my understanding of how discrete power amplifiers work. I wanted to learn by following along and performing several of the modifications in the videos. So building an LM3886 gainclone would not make more sense, because it would cost more for all the parts (higher voltage xformer, bigger heatsinks, bigger case, etc) and I would not have a convenient 10 part video training series to reference when testing and reviewing modifications. If I can get this amp to work without overheating, it will probably only produce 10 to 15 watt per channel, at most, when all is said and done. But the limited power does not concern me. I am just trying to keep the cost down (hopefully less than $100 or so all-in for transformer, case, kit, heatsink, transistors, etc).
So far:
$16 original OCL power amp kit
$25 Transformer (Triad)
$5 Replacement heatsink (1 ft section from a four ft long 1/8" x 2" aluminum angle)
$30 CU-622-A aluminum Bud box
$16 Four new replacement 2N3055 onsemi output transistors (Mouser)
$92 Total
At some point in the future, I might try building a higher powered gainclone type chip amp. But I am currently well stocked with several good sounding power amplifiers in the 60 to 250wpc range.
At some point in the future, I might try building a higher powered gainclone type chip amp. But I am currently well stocked with several good sounding power amplifiers in the 60 to 250wpc range.
That's where I'm at too. I don't listen to all the amps I currently have
I am curious to see if the genuine 2n3055s cure the overheating problem. I'll bet it will. Like Tony, I didn't see much difference between one channel with the stock 9012s & 9014s vs the other channel with known, proven subs bought from Mouser. They both got hot too quickly.
No problem. This is all good info. I noticed too, that after about 15 minutes or so, the outer two output transistors were peaking up into the 140F to 150F range while the inner two transistors were peaking up to about 110F to 130F. For some unknown reason the inner two power output transistors were running about 20F cooler.
After removing them, I did a diode check and they still check OK. No dead shorts or strange diode readings.
I'm using a cheap Sontax infrared thermometer that I picked up at Menards. I shine the red dot onto the transistor cases at a distance of about 12 inches. The reading bounces up and down quite a bit. When they get over 120F, they are hot enough to burn the tip of my index finger (ouch)!
I've heard those have a difficult time getting a reading off reflective surfaces. So it might bounce around because it is getting readings from the black lettering and it isn't realy possible to hold it perfectly in one spot by hand.
Aluminum angle and Bud box fully machined and ready for assembly. If the transistors overheat this time, it won't be due to insufficient sinking! The entire chassis will be a large 12 x 6 x 2.5" heatsink. The PCB will be held securely in place by the transistor mounting screws.
Before mounting, I set up my curve tracer again and measured the new 2N3055 transistors. I also re-measured the removed, overheating 3DD15D's to see how they compared. I used the same oscilloscope XY settings and curve tracer settings for the comparison. The four 2N3055's measured the same and seemed like a good matched set. Three of the four 3DD15D's, however, had a measured gain of about 5 times less than the 2N3055's! One of the 3DD15D's had a measured gain of about 3 times less that the 2N3055's. Houston, we have a problem!
Mounting the power output transistors was a very messy, time consuming job, involving several trips to the hardware store to get the right size screws, nuts, and spacers. So I thought I'd show a few pics of how it went.
The transistors came in a nice, padded, anti-static box. Mouser did a very good job of packaging and they arrived without bent pins. (unlike my Amazon order, arriving with bent over pins squished against other parts in a small plastic bag) :
I used small nylon spacers to keep the mounting screws from shorting against the heatsink bracket:
Then I applied a generous layer of heatsink grease onto the heatsink for the first transistor:
I then added a thin, clear mica insulator pad on top of the heatsink grease:
Then I applied a generous layer of heatsink grease to the transistor itself:
As I tightened the M3 x 10mm mounting screws, the grease compressed and squeezed out. I probably used a little too much grease, but too much is better than not enough.
Rinse and repeat for the remaining transistors and clean off the excess. Done:
Next, I mounted the aluminum angle bracket onto the Bud box bottom panel. I think this panel is actually the top of the "Converta" box, but I am using it as my bottom panel. The angle bracket is held in place by the back panel jacks, pots, switch, and IEC power inlet. And the 8 power output transistor mounting screws hold the PCB board firming in place with approximately 3/8" of clearance underneath the board.
Here is what the back panel looks like. I bolted the small heatsinks to the back panel, but there was only room enough for two of them. The other two will end up in my junk box:
That is where the project stands right now. I am getting ready to hook it back up to my bench power supply and give it the smoke test.
If you wanted to add more finned heat sinks you could install them inside the enclosure under the 2 pots by drilling a hole in the center - it might cut down on the protrusions back there as well . . .
Just a FYI, heatsink compound is best used sparingly. It sole purpose is to eliminate air in the pores/scratches of the metals. It's not as thermally conductive as metal, so any excess is actually hindering heat transfer.
You should apply a small amount, wiping in on, then use a razor blade to scrape the excess off.
Unfortunately Eric is 100% correct. You would have been better off using no compound and just the mica insulators. I would disassemble and redo it with the bare minimum compound.
Ok, I stand corrected. Thanks much for the grease application tip. This is one of the reasons I like to post detailed pictures and descriptions. To get good feedback like this. It is kind of a messy job, but I'll tear it apart and re-apply the grease properly. But before I do, I'll do a complete thermal power test and log the temperature data. I logged a detailed minute by minute temperature chart on the old transistors before disassembly. So I will be able to put together a spreadsheet comparison of the old verses the new transistors, mounted on the small and bigger heatsinks, with and without grease properly applied.
By the way, the amp passed the smoke test this morning. I also performed a short 10 minute thermal test at a low 1 watt/channel output level. No thermal runaway at all.
My application method for CPUs is to rub it on sparingly with a finger to both sides. Enough to basically leave a heavily smudged look on the surfaces. "Dirty" but not coated
Good to hear it's running nice and cool! I'll have to put some of those transistors on my next Mouser order.
Yeah, I hate thermal paste too. It seems to get everywhere you don't want it. I've gotten to the point of wearing gloves, but that doesn't keep it from getting on the desk.
I put together a spreadsheet of my thermal tests so far (attached). It looks like the new 2N3055 transistors and larger heatsinks are going to work out just fine. The spreadsheet tests were all done at 1.1 watts rms, but I just tested the new transistors for about 5 minutes at a power level of 10W rms per channel at 1kHz and the temperature only comes up another 10F or so to about 110F. This is running at the maximum power output just before clipping. When I get the entire chassis put together and raise it off the bench with rubber feet, the temperature should drop even more. Looks like this project is a go. Next up: Installing Tony's DC offset mod. When installed, this should increase power output to about 12wpc rms before clipping. 47mv is simply too much DC offset and is reducing the power output in the form of asymmetrical clipping.
@Tom_S said:
Good to hear it's running nice and cool! I'll have to put some of those transistors on my next Mouser order.
Yeah, I hate thermal paste too. It seems to get everywhere you don't want it. I've gotten to the point of wearing gloves, but that doesn't keep it from getting on the desk.
At this point, I'm not sure if it was the transistors or the heatsink that did the trick. When comparing the new and old transistors, there was a huge difference in gain on my curve tracer. Not sure what to make of this, because "in circuit" both sets of transistors produced about 40dB of closed loop gain. If I had to guess, I'd probably say that the transistors were the main culprit. If you watch Tony's 6 year old video carefully, you will notice that he did not seem to have the thermal runaway problem when using the smaller heatsinks.
@Steve_Lee said:
If you wanted to add more finned heat sinks you could install them inside the enclosure under the 2 pots by drilling a hole in the center - it might cut down on the protrusions back there as well . . .
There is not much room below the two pots. What type of finned heat sink would you suggest? The ones that came with the kit are a little too big to fit into this area inside.
I meant that you place the pot inside your existing heat sink using the pot's washer and nut to capture the assembly tight against the aluminum structure.
Comments
For a basic Gainclone, there used to be a ton of boards available. I may even have a set hanging around from Chipamps.com. I designed my own boards 2 decades ago, but I can't remember who I used - maybe PCBWay? I also used a surplus transformer from Apex Jr. It's a solid 55w/ch amp with no frills at all.
Yeah, I'm getting close to concluding that the Neurochrome 686 is going to be my reference amp when I get my reference speakers done (hopefully by SDC 2024).
That looks great Tom.
I bought pair of LM3886 chip amp kits from Aliexpress. Luckily mine showed up with real LM3886 chips, not the sometimes fake chips we heard stories about. They are bare bone circuit pcb's that don't have the Zobel or Theile circuits. Those are easily added. The basic pcb has a decent layout. I bought a 22-0-22 V Antek toroid transformer for them a few years ago and a bunch of PS caps from Apexjr. I should get off my rear and build a stereo amp with all that.
+1 on the Neurochrome amp.
Thanks for the gainclone feedback. The main reason I started this thread, however, was because xraytonyb recently released a 10 part video training series based on this cheap OCL kit. So I thought that it would be a good way to increase my understanding of how discrete power amplifiers work. I wanted to learn by following along and performing several of the modifications in the videos. So building an LM3886 gainclone would not make more sense, because it would cost more for all the parts (higher voltage xformer, bigger heatsinks, bigger case, etc) and I would not have a convenient 10 part video training series to reference when testing and reviewing modifications. If I can get this amp to work without overheating, it will probably only produce 10 to 15 watt per channel, at most, when all is said and done. But the limited power does not concern me. I am just trying to keep the cost down (hopefully less than $100 or so all-in for transformer, case, kit, heatsink, transistors, etc).
So far:
$16 original OCL power amp kit
$25 Transformer (Triad)
$5 Replacement heatsink (1 ft section from a four ft long 1/8" x 2" aluminum angle)
$30 CU-622-A aluminum Bud box
$16 Four new replacement 2N3055 onsemi output transistors (Mouser)
$92 Total
At some point in the future, I might try building a higher powered gainclone type chip amp. But I am currently well stocked with several good sounding power amplifiers in the 60 to 250wpc range.
That's where I'm at too. I don't listen to all the amps I currently have
Sorry for derailing the thread, Bill.
I am curious to see if the genuine 2n3055s cure the overheating problem. I'll bet it will. Like Tony, I didn't see much difference between one channel with the stock 9012s & 9014s vs the other channel with known, proven subs bought from Mouser. They both got hot too quickly.
No problem. This is all good info. I noticed too, that after about 15 minutes or so, the outer two output transistors were peaking up into the 140F to 150F range while the inner two transistors were peaking up to about 110F to 130F. For some unknown reason the inner two power output transistors were running about 20F cooler.
After removing them, I did a diode check and they still check OK. No dead shorts or strange diode readings.
I'm using a cheap Sontax infrared thermometer that I picked up at Menards. I shine the red dot onto the transistor cases at a distance of about 12 inches. The reading bounces up and down quite a bit. When they get over 120F, they are hot enough to burn the tip of my index finger (ouch)!
https://www.menards.com/main/tools/hand-tools/measuring-layout-tools/infrared-thermometers-moisture-meter/non-contact-infrared-thermometer/84850/p-1455229589746-c-1550760556438.htm
I've heard those have a difficult time getting a reading off reflective surfaces. So it might bounce around because it is getting readings from the black lettering and it isn't realy possible to hold it perfectly in one spot by hand.
Yes that is true. We do thermal imaging of electrical equipment and the surface finish has a significant impact.
Aluminum angle and Bud box fully machined and ready for assembly. If the transistors overheat this time, it won't be due to insufficient sinking! The entire chassis will be a large 12 x 6 x 2.5" heatsink. The PCB will be held securely in place by the transistor mounting screws.
What aluminum chassis is that? Dimensions?
Bud Converta box, model CU-622-A:
https://www.budind.com/product/general-use-boxes/converta-box-metal-electronics-box/cu-622-a/#group=series-products&external_dimensions_group=0&internal_dimensions=0
Nice machining! There's just something about seeing a nice spiral of aluminum swirling up as you drill with a good step-bit.
Before mounting, I set up my curve tracer again and measured the new 2N3055 transistors. I also re-measured the removed, overheating 3DD15D's to see how they compared. I used the same oscilloscope XY settings and curve tracer settings for the comparison. The four 2N3055's measured the same and seemed like a good matched set. Three of the four 3DD15D's, however, had a measured gain of about 5 times less than the 2N3055's! One of the 3DD15D's had a measured gain of about 3 times less that the 2N3055's. Houston, we have a problem!
2N3055: Sweep voltage: 40. Vertical Sensitivity: 5ma/division. Step Selector: 20ua current/step
3DD15D: Sweep voltage: 40. Vertical Sensitivity: 5ma/division. Step Selector: 20ua current/step
Wow! That's pretty telling.
I like tubes
Mounting the power output transistors was a very messy, time consuming job, involving several trips to the hardware store to get the right size screws, nuts, and spacers. So I thought I'd show a few pics of how it went.
The transistors came in a nice, padded, anti-static box. Mouser did a very good job of packaging and they arrived without bent pins. (unlike my Amazon order, arriving with bent over pins squished against other parts in a small plastic bag) :
I used small nylon spacers to keep the mounting screws from shorting against the heatsink bracket:
Then I applied a generous layer of heatsink grease onto the heatsink for the first transistor:
I then added a thin, clear mica insulator pad on top of the heatsink grease:
Then I applied a generous layer of heatsink grease to the transistor itself:
As I tightened the M3 x 10mm mounting screws, the grease compressed and squeezed out. I probably used a little too much grease, but too much is better than not enough.
Rinse and repeat for the remaining transistors and clean off the excess. Done:
Next, I mounted the aluminum angle bracket onto the Bud box bottom panel. I think this panel is actually the top of the "Converta" box, but I am using it as my bottom panel. The angle bracket is held in place by the back panel jacks, pots, switch, and IEC power inlet. And the 8 power output transistor mounting screws hold the PCB board firming in place with approximately 3/8" of clearance underneath the board.
Here is what the back panel looks like. I bolted the small heatsinks to the back panel, but there was only room enough for two of them. The other two will end up in my junk box:
That is where the project stands right now. I am getting ready to hook it back up to my bench power supply and give it the smoke test.
That's looking great! The added volume pots will come in handy too. Fingers crossed for cool running transistors!
If you wanted to add more finned heat sinks you could install them inside the enclosure under the 2 pots by drilling a hole in the center - it might cut down on the protrusions back there as well . . .
Just a FYI, heatsink compound is best used sparingly. It sole purpose is to eliminate air in the pores/scratches of the metals. It's not as thermally conductive as metal, so any excess is actually hindering heat transfer.
You should apply a small amount, wiping in on, then use a razor blade to scrape the excess off.
Unfortunately Eric is 100% correct. You would have been better off using no compound and just the mica insulators. I would disassemble and redo it with the bare minimum compound.
Ok, I stand corrected. Thanks much for the grease application tip. This is one of the reasons I like to post detailed pictures and descriptions. To get good feedback like this. It is kind of a messy job, but I'll tear it apart and re-apply the grease properly. But before I do, I'll do a complete thermal power test and log the temperature data. I logged a detailed minute by minute temperature chart on the old transistors before disassembly. So I will be able to put together a spreadsheet comparison of the old verses the new transistors, mounted on the small and bigger heatsinks, with and without grease properly applied.
By the way, the amp passed the smoke test this morning. I also performed a short 10 minute thermal test at a low 1 watt/channel output level. No thermal runaway at all.
My application method for CPUs is to rub it on sparingly with a finger to both sides. Enough to basically leave a heavily smudged look on the surfaces. "Dirty" but not coated
Good to hear it's running nice and cool! I'll have to put some of those transistors on my next Mouser order.
Yeah, I hate thermal paste too. It seems to get everywhere you don't want it. I've gotten to the point of wearing gloves, but that doesn't keep it from getting on the desk.
I put together a spreadsheet of my thermal tests so far (attached). It looks like the new 2N3055 transistors and larger heatsinks are going to work out just fine. The spreadsheet tests were all done at 1.1 watts rms, but I just tested the new transistors for about 5 minutes at a power level of 10W rms per channel at 1kHz and the temperature only comes up another 10F or so to about 110F. This is running at the maximum power output just before clipping. When I get the entire chassis put together and raise it off the bench with rubber feet, the temperature should drop even more. Looks like this project is a go. Next up: Installing Tony's DC offset mod. When installed, this should increase power output to about 12wpc rms before clipping. 47mv is simply too much DC offset and is reducing the power output in the form of asymmetrical clipping.
At this point, I'm not sure if it was the transistors or the heatsink that did the trick. When comparing the new and old transistors, there was a huge difference in gain on my curve tracer. Not sure what to make of this, because "in circuit" both sets of transistors produced about 40dB of closed loop gain. If I had to guess, I'd probably say that the transistors were the main culprit. If you watch Tony's 6 year old video carefully, you will notice that he did not seem to have the thermal runaway problem when using the smaller heatsinks.
There is not much room below the two pots. What type of finned heat sink would you suggest? The ones that came with the kit are a little too big to fit into this area inside.
I meant that you place the pot inside your existing heat sink using the pot's washer and nut to capture the assembly tight against the aluminum structure.