Tad. Thank you for your kind comments. I am much in favor of your suggestion that we make completely separate front end and output boards and not just one separable board. I was thinking about this also. Thank you for suggesting this to the group. Allowing people to buy separate front end and output boards is a very good idea indeed. I think that we need to pay close attention to very close spacing of components on the front end board so as not to inhibit the choice of components. We should be sure to leave adequate space for heat sinks for larger case transistors and capacitors. Your point about a robust connection between the front end board and output is a very good one. Jens is clearly the expert and professional when it comes to board layouts. Perhaps he would weigh in on this. ---HK
tryonziess said:
OnSemi might be interested in talking to you if you were planning to order 2-3K per month rather than a one shot deal. I tried it with TI for 1000+ OPA2134 and was told to call Digikey. It just isn't cost effective for them to deal with one shot customers for quantities about the same as a distributor would buy regularly.
Bob, The really good deals on output devices are usually several months lead time which makes for difficult assembly. Most of us are constructing on weekends and time stolen from lifes many necessary duties. I was just probing for a possible way to cut costs. Thanks for the response.
Just a little note -- I have been experimenting with the bias current on my Leach amp using Dr. Leach's method of bias adjustment. At this time the current is at 200ma. I was surprised at how much improvement I got in the sound. I still have not reached a point where the heatsinks get hot even warm. Any ideas on how far I can safely push the bias. Tad
Just a little note -- I have been experimenting with the bias current on my Leach amp using Dr. Leach's method of bias adjustment. At this time the current is at 200ma. I was surprised at how much improvement I got in the sound. I still have not reached a point where the heatsinks get hot even warm. Any ideas on how far I can safely push the bias. Tad
Yep it's frustrating having to wait for parts. It seems OnSemi only runs them once or twice a year. Back when I was trying to kit the PSU 2.2 it was really hard to find MJE1503x at any price.
Normal music doesn't generate a lot of heat, but at 200 mA (assuming total bias) and 60V rails, you're dissipating ~25W and should feel something.
How hot are your heat spreaders? If your sinks are staying cool, you might not be getting the heat to them. Check the temperature of the devices themselves and the spreader. If you didn't spend a lot of effort mating the spreader to the sinks you might not get good heat transfer even with the right amount of goop. Your heat sink may effectively consist of only the spreader.
Rule of thumb - around 20W per device, although I don't think you've got enough heat sink for that. Keep the heat sinks touchable for a few seconds - roughly 55C or less.
Or crunch the numbers for Class A operation ~.8C/w device to case (check data sheet) + 1.2 case to spreader through the insulator varies with type from .1 to 1.6), + another .7 or so spreader to heat sinks (varies wildly with surface matching). So you figure that the junction will be ~2.7C/W above the heat sink temperature.
Calculate how much power it will take to raise your heat sinks to 30C above ambient Power x C/W (each section rating / # of sections), and divide that by 10 devices and you know how much each dissipates and the resultant junction temperature. If that is above 100C, you need to limit the current.
It's a lot simpler than I made it sound. Check the Pass A75 article for a better explanation. Each thermal junction is like a resistance in series with the next and the heat is like a current. Multiple devices just put it all in parallel.
Anyone got a better rule for class AB heat sink sizing?
Normal music doesn't generate a lot of heat, but at 200 mA (assuming total bias) and 60V rails, you're dissipating ~25W and should feel something.
How hot are your heat spreaders? If your sinks are staying cool, you might not be getting the heat to them. Check the temperature of the devices themselves and the spreader. If you didn't spend a lot of effort mating the spreader to the sinks you might not get good heat transfer even with the right amount of goop. Your heat sink may effectively consist of only the spreader.
Rule of thumb - around 20W per device, although I don't think you've got enough heat sink for that. Keep the heat sinks touchable for a few seconds - roughly 55C or less.
Or crunch the numbers for Class A operation ~.8C/w device to case (check data sheet) + 1.2 case to spreader through the insulator varies with type from .1 to 1.6), + another .7 or so spreader to heat sinks (varies wildly with surface matching). So you figure that the junction will be ~2.7C/W above the heat sink temperature.
Calculate how much power it will take to raise your heat sinks to 30C above ambient Power x C/W (each section rating / # of sections), and divide that by 10 devices and you know how much each dissipates and the resultant junction temperature. If that is above 100C, you need to limit the current.
It's a lot simpler than I made it sound. Check the Pass A75 article for a better explanation. Each thermal junction is like a resistance in series with the next and the heat is like a current. Multiple devices just put it all in parallel.
Anyone got a better rule for class AB heat sink sizing?
Bob, Using my hand directly on the transistors I feel no warmth beyond ambient. With my infrared sensor the heatsink and the actual device are at 60 F. I also monitor the 3 device leads as they tie directly to the die. After running the amp for several hours -- and it sounds REAL good the temperature of the devices is no more than 65 F. I am driving 40 8 ohm speakers configured at 5 - 6 ohms total in series parallel. I am not by any means complaining it just seems strange to be running so cool. The devices are mounted with Berquist sil-pads and no goop. The heatsink contact area was mirror polished after I milled it and I lapped the backs of each transistor for good contact.
If increasing the bias and MONITORING the heat is an acceptable safe method I would think the end results would be better sound. I am not very knowledgeable here so I ask lots of questions. The success in this project to date just makes me want to try different things. Thanks Tad
If increasing the bias and MONITORING the heat is an acceptable safe method I would think the end results would be better sound. I am not very knowledgeable here so I ask lots of questions. The success in this project to date just makes me want to try different things. Thanks Tad
Glad to hear you lapped the spreader and sinks (now I remember you saying that earlier) - sounds like you're getting good heat transfer. and you should be safe increasing the bias. Cool sinks had me concerned. if your sinks are at 60F, your room must be COLD 
I measured Silpads at around 1.2C/W device to sink.
Bump the bias and measure is probably safer than counting on the datasheet rating of the sinks. Let it stabilize an hour or so keeping an eye on the temperature. As you increase the bias you should see that the devices get a bit hotter than the sinks.
Assuming that you get good transfer and spreading, I'd guess that you can get around 100W dissipation at idle. That's almost 1A of bias, which should get you 10-12W peak into your load in class A. You'll probably stay in class A for most listening.
Welcome to the class A club. Now you'll want to add even more capacitance to the power supply and regulate the front end.
I measured Silpads at around 1.2C/W device to sink.
Bump the bias and measure is probably safer than counting on the datasheet rating of the sinks. Let it stabilize an hour or so keeping an eye on the temperature. As you increase the bias you should see that the devices get a bit hotter than the sinks.
Assuming that you get good transfer and spreading, I'd guess that you can get around 100W dissipation at idle. That's almost 1A of bias, which should get you 10-12W peak into your load in class A. You'll probably stay in class A for most listening.
Welcome to the class A club. Now you'll want to add even more capacitance to the power supply and regulate the front end.
I am measuring the total bias at the input fuse on the board. I did a little more homework and found that my reading of 200ma is the total for all output devices. I have quite a way to go.
Now if I can find a really good frontend power supply design for this amp I would like to implement it into the next set of boards I construct. I was looking at Roender's shuntie frontend supply with a few more volts. Looks easy enough to put together. If I can configure it for 75 volts. The next set is dual mono with split supply.
I have found that those stiff large capacitance power supplys can be murder on my bridge rectifiers. They also eat up my variac fuses. I am going to add the current inrush protectors that came Friday from Digikey and see how that works.
This morning my old Mcintosh 2250 goes on fleabay. I have a better amp now. Then I can buy more stuff. Getting old has some rewards.
We might have taken our second GB a little past the designers new parameters. Will wait and see thanks Tad
Now if I can find a really good frontend power supply design for this amp I would like to implement it into the next set of boards I construct. I was looking at Roender's shuntie frontend supply with a few more volts. Looks easy enough to put together. If I can configure it for 75 volts. The next set is dual mono with split supply.
I have found that those stiff large capacitance power supplys can be murder on my bridge rectifiers. They also eat up my variac fuses. I am going to add the current inrush protectors that came Friday from Digikey and see how that works.
This morning my old Mcintosh 2250 goes on fleabay. I have a better amp now. Then I can buy more stuff. Getting old has some rewards.
We might have taken our second GB a little past the designers new parameters. Will wait and see thanks Tad
tryonziess said:
Not that far - you're probably going to end up at an amp total.
Haven't looked at Shuntie, but it shoudln't be too hard to configure for higher voltage with an eye to dissipation and device voltage ratings. Jens' PSU2.2 can be configured for 75V if you're willing to run a GB for more boards
You may find that you need a power resistor/relay soft start, especially once you go dual mono. I use a 6A slow blow with a 1KVA transformer and 180,000 uf and a CL60. A 10A fuse popped pretty regularly even with the CL60.
I think we were past them when we asked for a split board. While I can think of a whole list of wouldn't it be nice ifs, I would be perfectly happy with the original 10 output board or the rearranged one piece board.
Jens, you should get something that YOU want for your efforts. Thank you for making your design skill available to us lesser folk. Most of us are happy just to have a quality board available.
Bob, Once we get this design phase ironed out I will be more than willing to do a group buy -- if everyone was happy with the last one I handled -- including the psu 2.2.
My personal feelings on this design thing is that Jens be provided some free boards for his effort. That is small compensation for his knowledge, but let's be realistic here Jens is REAL good at this and his stuff works. My opinion anyway.
I would at this time be more than happy to build the original redesign Jens proposed with the added feedback network and other modifications. Like you stated we can wish for all kinds of new stuff, but someone else is doing the work on implementing it into a WORKABLE design. My 2 cents worth.
I have 1 amp per rail running now in the amp with no overheat problems. I will run it this way for awhile. Thanks to everyone on this forum for their time and knowledge in this project. Tad
My personal feelings on this design thing is that Jens be provided some free boards for his effort. That is small compensation for his knowledge, but let's be realistic here Jens is REAL good at this and his stuff works. My opinion anyway.
I would at this time be more than happy to build the original redesign Jens proposed with the added feedback network and other modifications. Like you stated we can wish for all kinds of new stuff, but someone else is doing the work on implementing it into a WORKABLE design. My 2 cents worth.
I have 1 amp per rail running now in the amp with no overheat problems. I will run it this way for awhile. Thanks to everyone on this forum for their time and knowledge in this project. Tad
Attached is a consolidated list which is accurate. The total shown on one of the wiki pages is incorrect. Any entries which are not numeric will not be counted. Who ever wishes to use this sheet to replace the two existing wiki lists is free to do so. I am not the owner of either of the lists ---HK
Thanks Jens, I didn't realize you'd updated it.
http://www.delta-audio.com/Leach_Clone/Leach_Amp_Clone_ver_0.4.pdf
http://www.delta-audio.com/Leach_Clone/Leach_Amp_Clone_ver_0.4.pdf
would anybody be nice and reply to my previous post
on that post I am talking to the pcbs at http://www.diyaudio.com/forums/showthread.php?postid=1387621#post1387621
now http://www.diyaudio.com/forums/showthread.php?postid=1402369#post1402369 I understand that there are new pcbs which are separate
are these pcbs (older or new ones) only prototypes or ready to be purchased
and please tell me who I have to call to include me on the ordering page and what Is phe price
on that post I am talking to the pcbs at http://www.diyaudio.com/forums/showthread.php?postid=1387621#post1387621
now http://www.diyaudio.com/forums/showthread.php?postid=1402369#post1402369 I understand that there are new pcbs which are separate
are these pcbs (older or new ones) only prototypes or ready to be purchased
and please tell me who I have to call to include me on the ordering page and what Is phe price