Hi Kaplaars,
One thing to bear in mind (and ignore this part if you know it) is that running the transistors hot actually reduces their capacity, ie the TO3 transistor that dissipates 200W can actually only do so at a temp of 20c, by 100c it is, for example, only a 100W transistor. This is called thermal derating and needs to be considered before choosing final settings.
Safe has a wide variety of definitions, it really depends on your assumptions and expectations. Are you talking about whether the amp survives a short on the output? or whether someone that stumbles and puts a hand on the amp then needs to go to hospital? Or what if there is a persistent very low impedance load (ask how I know ;-) on the output what happens when you continue to drive the amp, increasing the input to compensate for the apparent lack of volume...
I don't think I have a clear understanding of your goals, at one point you make it clear you don't need a lot of class A bias because you don't use that much power, later (IIUC) you are figuring the maximum bias the output stage, mounted on your heatsinks can dissipate before failing.
Stuart
One thing to bear in mind (and ignore this part if you know it) is that running the transistors hot actually reduces their capacity, ie the TO3 transistor that dissipates 200W can actually only do so at a temp of 20c, by 100c it is, for example, only a 100W transistor. This is called thermal derating and needs to be considered before choosing final settings.
Safe has a wide variety of definitions, it really depends on your assumptions and expectations. Are you talking about whether the amp survives a short on the output? or whether someone that stumbles and puts a hand on the amp then needs to go to hospital? Or what if there is a persistent very low impedance load (ask how I know ;-) on the output what happens when you continue to drive the amp, increasing the input to compensate for the apparent lack of volume...
I don't think I have a clear understanding of your goals, at one point you make it clear you don't need a lot of class A bias because you don't use that much power, later (IIUC) you are figuring the maximum bias the output stage, mounted on your heatsinks can dissipate before failing.
Stuart
The last build I did was just after my daughter was born, so it was important to me that the amps be 'safe' in the domestic sense, ie
As as example of safety choices made in building a monoblock, fan cooled KSA50(75?):
- We used a normal 5A fuse and CL30 on the input to the 625VA transformer.
- We used resetting 85c cut-offs on the heatsinks, to protect against fan failures etc.
- We added basic speaker protection (on/off thump and DC offset detection).
The final temperature of the outside of the amp was determined by how fast we wanted to run the fans...we had the top and bottom plates remachined after we realised the fans were running faster (and noisier) than we liked in order to move the required air to keep the transistors at a reasonable temperature...
We kept the outside of the amplifier cases to ~45c and the heatsinks to 65, transistors were more like 95, IIRC. It took a number of iterations of settings and we didn't finish winding and potting the transformers until we were pretty clear on what the overall dissipation parameters could be (~200w, ~40V, ~2.5A) giving 100W class A / 4ohms.
HTH
Stuart
As as example of safety choices made in building a monoblock, fan cooled KSA50(75?):
- We used a normal 5A fuse and CL30 on the input to the 625VA transformer.
- We used resetting 85c cut-offs on the heatsinks, to protect against fan failures etc.
- We added basic speaker protection (on/off thump and DC offset detection).
The final temperature of the outside of the amp was determined by how fast we wanted to run the fans...we had the top and bottom plates remachined after we realised the fans were running faster (and noisier) than we liked in order to move the required air to keep the transistors at a reasonable temperature...
We kept the outside of the amplifier cases to ~45c and the heatsinks to 65, transistors were more like 95, IIRC. It took a number of iterations of settings and we didn't finish winding and potting the transformers until we were pretty clear on what the overall dissipation parameters could be (~200w, ~40V, ~2.5A) giving 100W class A / 4ohms.
HTH
Stuart
Stuart, I am sorry that I went with the 40v/1000kva transfos. They were on sale for $65 each--12v/1A and 18v/1A secondaries included. I went to the metal scrap yard and found really nice very heavy irregular shaped extruded aluminum where I will cut diagonal slots in the direction of the incoming air. Since this is my next step, my current concern is about making the heat sinks live to the TO3 transistors and is there a way to side step this setup. After understanding this, I think I will fire this baby up--with fire extinguisher not too far away.
So, we are supporting people here who are ripping off the hard work and reputation of legit' companies? Jims does good work. Usually the design is slightly modified. But he is still ripping off Krell, Music Fidelity, and others. It cost these companies a lot of money to develop the products and build their reputation. Heck, I'd love to have a KAS 100. Intellectual property theft is, well just plain theft. I don't care it is a 20 year old design. I want to see Jims stand up on their own. Build a board "inspired by" not ripped off. I know they can. We just have to stop buying stolen designs.
Hi Stuart,
Thanks for your reply! Appreciate it
Yes, knew that. That is why I got worried because, temperature is pretty high. I had to check and wanted to know if it was OK. Off course, I want, and have to be operating in the SOA area. I am pretty curious to what temperature the original KSA-100 is operating. I almost can't believe it keeps temperature below 70 degrees with 0.625 mV bias.
My goals are fun and to learn But the tecnical goals are a bit variable I want to reach for the stars, but there are a lot of different, unknown, parameters involved. I have discovered most, but some are still unknown. One of this parameters was for example C/W value from the heatsinks. The heatsinks are surplus, without documentation. Therefore I had to determine the C/W-value. Now I have a rough estimate I can adjust bias to it. I think I will try to bias it to a KSA-50, so 442mV across the 1 ohm resistors. The PSU is capable of delivering much more current, but
What I did for safety are:
I've did some further calculations, and yes, made a few mistakes in the post above
My calculation of C/W seems to be a good estimate. When having an ambient temperature of 30 degrees and using 270 mV bias, the temperature would theoretically become: Tambient + C/W * bias * rail voltage * output devices --> 30 + 0.27 * 0.270 * 51 * 8 = 60 degrees @ the heatsinks. I think temperature is even a bit lower. But can hold my hands without a problem to the theatsinks, so it is most certainly not warmer than 60 degrees. So great news, the heatsinks actually do their job pretty well.
Tj would in this situation be: Theatsink + Tr * power dissipated --> 60 + 0,7 * 51 * 0,270 = 69,6 degrees. This is worst case scenario, temperatures are seldom higher than 30 degrees here in Holland and Tr is maximun value. Also I my calculations are without load, so when applying a load, there is a good chance temperature is a few degrees lower at my trannies.
I think I shift my goal to biassing it to KSA-50 bias, which is 444 mV. That would be more than enough, and I don't have to ruin my design So same calculation would give 30 + 0.27 * 0.444 * 51 * 8 = 78,9 degrees on the heatsinks, and Tj would be 78,9 + 0,7 * 51 * 0,444 = 94,75 degrees. I could not find a SOA-curve of the MJ15003/4 to check if they could deliver the current needed when operating at this temperature.
625 mV is too tricky: 30 + 0.27 * 0.625 * 51 * 8 = 98,9 degrees on the heatsinks, and Tj would be 98,9 + 0,7 * 51 * 0,444 = 121.2.
Hope my calculations are right
Thanks for your reply! Appreciate it
Yes, knew that. That is why I got worried because, temperature is pretty high. I had to check and wanted to know if it was OK. Off course, I want, and have to be operating in the SOA area. I am pretty curious to what temperature the original KSA-100 is operating. I almost can't believe it keeps temperature below 70 degrees with 0.625 mV bias.My goals are fun and to learn
But the tecnical goals are a bit variable I want to reach for the stars, but there are a lot of different, unknown, parameters involved. I have discovered most, but some are still unknown. One of this parameters was for example C/W value from the heatsinks. The heatsinks are surplus, without documentation. Therefore I had to determine the C/W-value. Now I have a rough estimate I can adjust bias to it. I think I will try to bias it to a KSA-50, so 442mV across the 1 ohm resistors. The PSU is capable of delivering much more current, but What I did for safety are:
- one main fuse for when the main leads go short circuit.
- 4 secondairy fuses for when outputs go short circuit.
- speaker protection (DC offset, on / off thump).
- an other fuse with the primaries of the little transformer which is mounted at the powerboard for the speaker protection.
- two thermostats (rated 75 degrees)
- heavy 10 0hm / 10 W (replaced the previous used, lighter ones compairing to previous pictures) circuit breakers, so when there is a short circuit to GND, they don't act as fuse..
- soft start PCB. Next to delaying power on, it switches the amp on with two heavy relays. This way I dont have to use a heavy switch (which can more easily fail than the relays) but a little momentary switch instead.
I've did some further calculations, and yes, made a few mistakes in the post above
My calculation of C/W seems to be a good estimate. When having an ambient temperature of 30 degrees and using 270 mV bias, the temperature would theoretically become: Tambient + C/W * bias * rail voltage * output devices --> 30 + 0.27 * 0.270 * 51 * 8 = 60 degrees @ the heatsinks. I think temperature is even a bit lower. But can hold my hands without a problem to the theatsinks, so it is most certainly not warmer than 60 degrees. So great news, the heatsinks actually do their job pretty well.
Tj would in this situation be: Theatsink + Tr * power dissipated --> 60 + 0,7 * 51 * 0,270 = 69,6 degrees. This is worst case scenario, temperatures are seldom higher than 30 degrees here in Holland and Tr is maximun value. Also I my calculations are without load, so when applying a load, there is a good chance temperature is a few degrees lower at my trannies.
I think I shift my goal to biassing it to KSA-50 bias, which is 444 mV. That would be more than enough, and I don't have to ruin my design
So same calculation would give 30 + 0.27 * 0.444 * 51 * 8 = 78,9 degrees on the heatsinks, and Tj would be 78,9 + 0,7 * 51 * 0,444 = 94,75 degrees. I could not find a SOA-curve of the MJ15003/4 to check if they could deliver the current needed when operating at this temperature.625 mV is too tricky: 30 + 0.27 * 0.625 * 51 * 8 = 98,9 degrees on the heatsinks, and Tj would be 98,9 + 0,7 * 51 * 0,444 = 121.2.
Hope my calculations are right
I can tell you for sure your driver heat sink is way too small. Kaplaars is almost large enough. Those drivers run steamy hot when the amp is biased correctly andthey are actually an excellent sounding small amplifier in their own right. When I tested the origonal prototype boards about 5 years ago I found this out really quick...
Last edited:
Ai Mark, I hope that Spurtles heat sinks will do. Mine stay very cool. Total dissipation is about 15 W.
I made by the way a little spreadsheet with Excel. It calculates automaticly if your trannies are operating in SOA You can download it from my ftp: http://michaelq.home.xs4all.nl/Projects/Bias vs. power derating.xlsx
Hope it could help somebody. I found out that even at 625 mV bias, I am still on the safe side, max power handling is still 113 W per device (if my spreadsheet is OK). When operating at 90 degrees Celsius, would that shorten lifespan of the trannies? It feels unnatural when operating at temperatures that high, have no intention to do that. But I am curious. From what I can make up of the datasheets, it should not be a problem.
I made by the way a little spreadsheet with Excel. It calculates automaticly if your trannies are operating in SOA
You can download it from my ftp: http://michaelq.home.xs4all.nl/Projects/Bias vs. power derating.xlsxHope it could help somebody. I found out that even at 625 mV bias, I am still on the safe side, max power handling is still 113 W per device (if my spreadsheet is OK). When operating at 90 degrees Celsius, would that shorten lifespan of the trannies? It feels unnatural when operating at temperatures that high, have no intention to do that. But I am curious. From what I can make up of the datasheets, it should not be a problem.
heat sinks
Which heat sinks are you referring to? MJE15032/3? Klapaars and I had communicated earlier about these heatsinks and agreed the MJEs did not dissipate too much heat and the heatsinks I have will do the job. I plan to have the power supply in it's own case therefore seperating its heat from the amplifier's circuitry. If computers CPUs can use water cooling (I see r-tools.com program considers water cooling), why isn't it possible to cool off TO3s this way? I have seen whole working mother boards submerged in a fish tank with a circulating bath of some special liquid of some sort here in Chicago at a local computer shop I must get a picture of it for reference, but the liquid cooling type I am speaking of is are those attached directly to the computer CPU. What a mouthful.
That would be very VERY cool David! I've never seen a water cooled KSA before. It certainly could work, but I think you will have to fabricate some pieces to attach the watercooling and TO-3's to. I don't know if there are DIY TO-3 kits, never seen one.
Hmm when rethinking about the little heatsink you've mounted to the MJ15032/33's, Mark could have a point David. The heatsink I use stays pretty cool, but is not fully biassed right now and off course a lot larger. I would say my heatsink is even a bit to large. From the other hand.... I have not tried what temperature will do when fully biassed. When I check Wikipedia, I see temperatures in Chicago are a bit like here in Holland. You could always try what it will do. Tj is for MJ15032/3 150 degrees C (yes I am all into thermo related stuff right now
), so even at let say 60 degrees, there should not be a real problem but offcourse the cooler, the better.Michael, I suggest that you borrow from someone thermometer and measure temperature on the heatsinks exactly. I was thinking that the temperature is too high on my amp when I touched with my fingers, than it really was. I measured with a thermometer. 50 degrees is pretty hot to the touch, and at 55 degrees you can hold your fingers only a few seconds.
To quote N. Pass: "Human skin has the remarkable characteristic that we think 40° is comfortable, 45° is hot, 50° is very hot, and 55° is untouchable"
About Dale resistors, I told you so...I'm not surprised, but you can now sleep peacefully
To quote N. Pass: "Human skin has the remarkable characteristic that we think 40° is comfortable, 45° is hot, 50° is very hot, and 55° is untouchable"
About Dale resistors, I told you so...I'm not surprised, but you can now sleep peacefully
I have at least one (maybe two) water cooled to3 plate(s), hard to call it a heatsink really, it has no fins, just a drilled plate and a water pipe soldered or brazed into the surface.
If it were something someone wanted to try I could probably be convinced to track it down...I'm visiting family in England atm, but I should be back at home at the end of august.
If it were something someone wanted to try I could probably be convinced to track it down...I'm visiting family in England atm, but I should be back at home at the end of august.
The immersed PC display I saw used mineral oil, IIUC it's not hydroscopic and largely inert as far as other components are concerned. If you are going to immerse the whole motherboard you can probably just dispense with anything more than trivial heat-sinks with wide space fins, oil and water are much harder to pump between closely spaced fins.
Water cooling is an excellent idea. Just be sure to use the proper type of coolant or you might end up with mold growing in the circulation system. I've seen this happen in a Series 1 Barco D-Cinema projector where the factory coolant was replaced with antifreeze from Wal-Mart. Newer Series 2 D-Cinema stuff are all sealed type of cooling system and this won't happen. If you can use copper for your heat sinks and silver solder copper tubing to that you can make up fairly inexpensive custom heat sinks from 1/4" copper sheet.
Mark
Mark
Hahaha your comments always comforts me Dean Will do. Unfortunatelly lab @university is closed, so I can't borrow one at the moment :-(.
@John, you could use de-ionized water, water itself is not conductive, but electrolytes in it like Na+, Ca2+, CO32- and so on cause the conductive character from water. Another advantage of de-ionised water is that it almost does not cause corrosion. The more electrolytes water contains, the faster corrosion can occur (that is why boats have to be protected from corrosion by mounting zinc-blocs underneath because of RedOx processes). You can also use demineralized oil, which is not conductive either. But I would say water is more favored, it's heat capacity is higher, so it can absorb more heat (joules) per kg than oil when augmenting 1 degree in temperature (oil ~2KJ / kg * K, water ~4KJ /kg * K).
I have some news too, am afraid that it is not positive. Well start positive first; I have bought a very nice Audio Generator from Peaktech! I should have had this device far more earlier, this works so much better than using an computer, and I can do square wave tests.
So I did some tests, used 8 ohm dummy load, did not connect a capacitor parallel, will do that later. I first want to verify if this results are OK. Results on my scope are exactely the same for both channels. Bias was 450mV. You can see VAC RMS on the right meter below the scope.
1. 400Hz, sine wave, max output (30 VAC RMS @8 ohm):
2. 1000Hz, sine wave, max output (29.5 VAC RMS @8 ohm):
3. 400Hz square wave, output 1VAC RMS
4. 1000Hz square wave, output 1VAC RMS
5. 10 kHz square wave, output 1VAC RMS
So far so good, but at max power, there is a BIG overshoot, which worries me
6. 400Hz square wave, max output
7. 1000Hz square wave, max output
8. 10 kHz square wave, max output
Fortunatelly I don't see any oscillation so far, just overshoot. Tried with and without zobel, but still overshoot at max output. I can think of a few causes:
- The amp clips at this amount of power (when switching to sine wave you can see it clips).
- The VAC meter which is connected to the scope it has a 100:1 devider inside, so when I measure 100VAC, it sends 1VAC to the scope.
- I've made a mistake, the amp is faulty.
- It is common and not a problem.
- I don't know what I am doing
Guys, need the expertise again. Can someone judge my results?
Will do. Unfortunatelly lab @university is closed, so I can't borrow one at the moment :-(. @John, you could use de-ionized water, water itself is not conductive, but electrolytes in it like Na+, Ca2+, CO32- and so on cause the conductive character from water. Another advantage of de-ionised water is that it almost does not cause corrosion. The more electrolytes water contains, the faster corrosion can occur (that is why boats have to be protected from corrosion by mounting zinc-blocs underneath because of RedOx processes). You can also use demineralized oil, which is not conductive either. But I would say water is more favored, it's heat capacity is higher, so it can absorb more heat (joules) per kg than oil when augmenting 1 degree in temperature (oil ~2KJ / kg * K, water ~4KJ /kg * K).
I have some news too, am afraid that it is not positive. Well start positive first; I have bought a very nice Audio Generator from Peaktech! I should have had this device far more earlier, this works so much better than using an computer, and I can do square wave tests.
So I did some tests, used 8 ohm dummy load, did not connect a capacitor parallel, will do that later. I first want to verify if this results are OK. Results on my scope are exactely the same for both channels. Bias was 450mV. You can see VAC RMS on the right meter below the scope.
1. 400Hz, sine wave, max output (30 VAC RMS @8 ohm):
2. 1000Hz, sine wave, max output (29.5 VAC RMS @8 ohm):
3. 400Hz square wave, output 1VAC RMS
4. 1000Hz square wave, output 1VAC RMS
5. 10 kHz square wave, output 1VAC RMS
So far so good, but at max power, there is a BIG overshoot, which worries me
6. 400Hz square wave, max output
7. 1000Hz square wave, max output
8. 10 kHz square wave, max output
Fortunatelly I don't see any oscillation so far, just overshoot. Tried with and without zobel, but still overshoot at max output. I can think of a few causes:
- The amp clips at this amount of power (when switching to sine wave you can see it clips).
- The VAC meter which is connected to the scope it has a 100:1 devider inside, so when I measure 100VAC, it sends 1VAC to the scope.
- I've made a mistake, the amp is faulty.
- It is common and not a problem.
- I don't know what I am doing
Guys, need the expertise again. Can someone judge my results?
I have to learn to check, check, double check before worrying haha. Yes friends, problem solved. I was allready surprised that my RMS VAC meter gave 30 VAC RMS, I've measured 40 VAC RMS before. Turns out that the measuring clips made a bad contact which caused the problems. I realy hate those clips, my next investment will be some relieable Hirschmann clips.
New picture, 10 kHz square wave:
I think this is more like it
Yes friends, problem solved. I was allready surprised that my RMS VAC meter gave 30 VAC RMS, I've measured 40 VAC RMS before. Turns out that the measuring clips made a bad contact which caused the problems. I realy hate those clips, my next investment will be some relieable Hirschmann clips.New picture, 10 kHz square wave:
An externally hosted image should be here but it was not working when we last tested it.
I think this is more like it
Sorry for kicking the topic, but I could not edit the post above.
Well guys, I blamed the clips, because when changing them the problems were over. But, believe it or not, but 4 minutes after I sended the post above, my oscilloscope went blank and loads of smoke came out of it. It was pretty impressive. Pfew, that stinks. Took the oscilloscope outside. It kept smoking for a wile.
I had an earlier discussion with David (Spurtle) about the so called 'rattle capacitor'. These oscilloscopes have one too. They are from Rifa. Altough quality is not that bad, after 30 years they go short circuit (X2 capacitor my a**). Here is a picture of it, I would not even imagine what would have been happened if I don't was in my room:
(blue one is the replacement)
Unbelievable how much smoke can come out of such a little capacitor. Replaced it, cleaned the print and decided to clean the potentiometers and switches with some Tuner 600 spray. This is very good spray btw, leaves absolutely no residue. Prayed that it would work again, because Kaplaars is a bit broke nowadays And IT DID!!!
The oscilloscope had some stability problems earlier with triggering and al sorts of strange distortions. Thought that was due to the clips, but it turned out it was actually caused by the oscilloscope itselve! So another note to myself, when testing equipment NEVER leave the room.
Well guys, I blamed the clips, because when changing them the problems were over. But, believe it or not, but 4 minutes after I sended the post above, my oscilloscope went blank and loads of smoke came out of it. It was pretty impressive
. Pfew, that stinks. Took the oscilloscope outside. It kept smoking for a wile.I had an earlier discussion with David (Spurtle) about the so called 'rattle capacitor'. These oscilloscopes have one too. They are from Rifa. Altough quality is not that bad, after 30 years they go short circuit (X2 capacitor my a**). Here is a picture of it, I would not even imagine what would have been happened if I don't was in my room:
An externally hosted image should be here but it was not working when we last tested it.
(blue one is the replacement)
An externally hosted image should be here but it was not working when we last tested it.
Unbelievable how much smoke can come out of such a little capacitor. Replaced it, cleaned the print and decided to clean the potentiometers and switches with some Tuner 600 spray. This is very good spray btw, leaves absolutely no residue. Prayed that it would work again, because Kaplaars is a bit broke nowadays
And IT DID!!!The oscilloscope had some stability problems earlier with triggering and al sorts of strange distortions. Thought that was due to the clips, but it turned out it was actually caused by the oscilloscope itselve! So another note to myself, when testing equipment NEVER leave the room.
Water cooling
Fantastic. I have a scrap metal yard near my home in Chicago that deals mostly with aluminum and copper (money makers). I wanted to use a car heater radiator (the heater core which is usually located below the dash and if the heater core can heat the interior of a car in dead winter, it should be perfect to dissipate the heat of the TO3s), with electric fans, parallel fluid lines attached to the copper plates holding TO3s and a temperature sensing circuitry that would control fan speed and circulating fluids--no different than that that is used in modern vehicles. Imaging, no need for filtering forced air into the amplifier's cabinet. Hey, it's just an idea and a project for the coming winter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.