I'm planning on building a very high power amp some time in the near future, and amp trying to settle on a design.
Right now, I have two of these
1.9kVA trafos. I also have 4 x 22mF 100V caps, and 4 x 10mF 100V caps.
I also have a bunch (30 of each) of OnSemi MJE4281/4302 BJTs, so the design needs to use those.
Given the size of the PS, and the amount of output transistors I can thow at it, I'm looking for something that puts out around 2kW or more at the lowest stable impedance. Since it'll be (obivously
) a sub amp, super high fidelity isn't the most important criteria, efficiency is moreso.
Right now I'm leaning to a slightly modded ESP P68. Is there anything else out that that would be good to look at before I put an order in with Rod?
Right now, I have two of these
1.9kVA trafos. I also have 4 x 22mF 100V caps, and 4 x 10mF 100V caps.
I also have a bunch (30 of each) of OnSemi MJE4281/4302 BJTs, so the design needs to use those.
Given the size of the PS, and the amount of output transistors I can thow at it, I'm looking for something that puts out around 2kW or more at the lowest stable impedance. Since it'll be (obivously
) a sub amp, super high fidelity isn't the most important criteria, efficiency is moreso. Right now I'm leaning to a slightly modded ESP P68. Is there anything else out that that would be good to look at before I put an order in with Rod?
Are they 208V primary and 120V secondary as stated in the auction? Then it would be difficult to hook them up to standard mains voltages... And the secondary voltage is somewhat high... Adding/removing windings is nearly impossible because you cannot remove the resin without damaging the transformer.
These are good for 80..85V rail voltage... Quite a common number for high power audio amplifiers. But won't get the power you want out of these rails unless you go bridge mode.
Very good devices.
I think if you get new transformers and build two pretty standard +/- 85V amplifiers, using them in bridge-mode, your goals are achievable.
Oh, I see. You meant MJL4281. 230W 15A devices in plastic cases. Power must be derated at 1.84W/C.
As has been stated already, you won't et 2kW output into 8-ohms without bridging. In single-ended (not bridged) the most you'll get is approx. 500W into 8-ohms, 1000W into 4 and 2000W into 2.
Bridging is probably the most effective way to get the max reading on the Richter scale. You'd need to make sure the P68 is capable of this or not and whether it has adequate fault protection (I am not familiar with the design). Bridging with an 8-ohm subwoofer will give you 2000W.
If you bridge using an 8-ohm nominal sub-woofer and you want good reliability it is prudent to design the amp to withstand a 4-ohm resistive load. In this case, and assuming a class AB with low bias current, the average dissipation of all the transistors can be as high as 1700W. That is the total for both amps in bridge mode. Assuming your heatsinks/fans can keep the transistor cases at no more than 70C, each transistor will withstand 147W. Ergo, you need at least 1700/147=12 transistors. But the peak dissipation is the limiting factor: with 12 devices they will each suffer 360W momentarily - not ideal. Best to get this below 230W in which case you'll need 20 devices in total. That's 5 per rail per amp.
In summary, you can reliably drive an 8-ohm nominal subwoofer in bridge mode up to 2000W if you use 20 of your transistors and keep them at 70C or lower. This is a conservative calculation. You could use less but I would strongly recommend power limiting circuitry to protect the transistors from exceeding their SOAs. In any case, you should have some protection against accidentally shorting the outputs and a thermal switch on the heatsinks to cut-out at 70C.
Whether or not you can adapt the P68 to this end is unknown to me.
As has been stated already, you won't et 2kW output into 8-ohms without bridging. In single-ended (not bridged) the most you'll get is approx. 500W into 8-ohms, 1000W into 4 and 2000W into 2.
Bridging is probably the most effective way to get the max reading on the Richter scale. You'd need to make sure the P68 is capable of this or not and whether it has adequate fault protection (I am not familiar with the design). Bridging with an 8-ohm subwoofer will give you 2000W.
If you bridge using an 8-ohm nominal sub-woofer and you want good reliability it is prudent to design the amp to withstand a 4-ohm resistive load. In this case, and assuming a class AB with low bias current, the average dissipation of all the transistors can be as high as 1700W. That is the total for both amps in bridge mode. Assuming your heatsinks/fans can keep the transistor cases at no more than 70C, each transistor will withstand 147W. Ergo, you need at least 1700/147=12 transistors. But the peak dissipation is the limiting factor: with 12 devices they will each suffer 360W momentarily - not ideal. Best to get this below 230W in which case you'll need 20 devices in total. That's 5 per rail per amp.
In summary, you can reliably drive an 8-ohm nominal subwoofer in bridge mode up to 2000W if you use 20 of your transistors and keep them at 70C or lower. This is a conservative calculation. You could use less but I would strongly recommend power limiting circuitry to protect the transistors from exceeding their SOAs. In any case, you should have some protection against accidentally shorting the outputs and a thermal switch on the heatsinks to cut-out at 70C.
Whether or not you can adapt the P68 to this end is unknown to me.
Sorry, it's the MJLs.
Given the power of this amp and it's use for subs, I'd have no problem with it putting out max power at 2 ohms, with 500W or so at 8 ohms. Unless I'm going with a small box Tumult in a Linkwitz tranform or something, I'd never use it on just one sub anyway.
Also, as it is a sub amp, some measure of distortion is not a big deal. Even a class b output stage is fine. The biggest thing is efficiency. Given the large amount of power, even a small increase in efficiency would translate into a large savings in heatsinking needs.
Given the power of this amp and it's use for subs, I'd have no problem with it putting out max power at 2 ohms, with 500W or so at 8 ohms. Unless I'm going with a small box Tumult in a Linkwitz tranform or something, I'd never use it on just one sub anyway.
Also, as it is a sub amp, some measure of distortion is not a big deal. Even a class b output stage is fine. The biggest thing is efficiency. Given the large amount of power, even a small increase in efficiency would translate into a large savings in heatsinking needs.
In terms of efficiency your best class AB amp will do 71% with worst-case output amplitude. Note that the worst efficiency does not occur at maximum output power.
Rule of thumb: to work out the approx. power dissipation of your electronics, take the max average speaker power and multiply by 3/7.
Eg: 2000W max average speaker power => at least 870W of heat to get rid of in your electronics (depending upon efficiency).
Rule of thumb: to work out the approx. power dissipation of your electronics, take the max average speaker power and multiply by 3/7.
Eg: 2000W max average speaker power => at least 870W of heat to get rid of in your electronics (depending upon efficiency).
All of the input circuitry would be done on a PCB, but I'd hardwire everything past the drivers, much like the ampslab products.
I'm thinking for heatsinking using this sink from the Canadian group buy, if it ever goes through.
Heatsink
Use 18-24" of it, and include some 120mm fans to keep it cool.
An externally hosted image should be here but it was not working when we last tested it.
I'm thinking for heatsinking using this sink from the Canadian group buy, if it ever goes through.
Heatsink
Use 18-24" of it, and include some 120mm fans to keep it cool.
What about the Leach Superamp?
Is there any reason why it couldn't be used with the 4302/4281 instead of the 15003/15004?
I would think the class G Leach would give better efficiency than the ESP P68
Is there any reason why it couldn't be used with the 4302/4281 instead of the 15003/15004?
I would think the class G Leach would give better efficiency than the ESP P68
"What about the Leach Superamp? "
Nice amplifier, Heathkit sold it as the AA1800.
"Is there any reason why it couldn't be used with the 4302/4281 instead of the 15003/15004?"
They would be OK with more devices and some base resistors.
"I would think the class G Leach would give better efficiency than the ESP P68 "
It would be quite easy to make the Leach Superamp class G if you want to, as is they are class AB.
Nice amplifier, Heathkit sold it as the AA1800.
"Is there any reason why it couldn't be used with the 4302/4281 instead of the 15003/15004?"
They would be OK with more devices and some base resistors.
"I would think the class G Leach would give better efficiency than the ESP P68 "
It would be quite easy to make the Leach Superamp class G if you want to, as is they are class AB.
Get the TIFF viewer
http://www.alternatiff.com/
http://patft.uspto.gov/netacgi/nph-...,387,876.WKU.&OS=PN/5,387,876&RS=PN/5,387,876
click on 'images'
http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
The $6 Nelson Pass A40 board makes a nice front end, single ended diff input and VAS with active current sources, bias network.
http://www.audioxpress.com/bksprods/pcbs/solidstate.htm
The $12 Erno Borberly EB60 board is similar to the Leach, cross-coupled dual diff inputs, complementary VAS, bias network.
http://www.alternatiff.com/
http://patft.uspto.gov/netacgi/nph-...,387,876.WKU.&OS=PN/5,387,876&RS=PN/5,387,876
click on 'images'
http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
The $6 Nelson Pass A40 board makes a nice front end, single ended diff input and VAS with active current sources, bias network.
http://www.audioxpress.com/bksprods/pcbs/solidstate.htm
The $12 Erno Borberly EB60 board is similar to the Leach, cross-coupled dual diff inputs, complementary VAS, bias network.
Ok, after thinking about it I've decided my best bet, based on what I currently have, would be to have two ESP P68s in the same chassis, with a selector switch allowing bridged or stereo operation.
With +/- 94V rails, I should get about 1000W into 4 ohms. 8 devices per rail should be more than enough, I'd think. That's 16 transistors per amp.
For heatsinks, I was thinking of using two of these...
They're 356mm x 150mm x 80mm. I'd place them with the fins touching and the mounting surfaces on opposite ends, so giving a 150mm x 160mm closed channel. Then, on the ends of the case, I'd run 120mm intake and an exhaust fans.
Sound adequate? Should be able to give 2kW with plenty of headroom.
With +/- 94V rails, I should get about 1000W into 4 ohms. 8 devices per rail should be more than enough, I'd think. That's 16 transistors per amp.
For heatsinks, I was thinking of using two of these...
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
They're 356mm x 150mm x 80mm. I'd place them with the fins touching and the mounting surfaces on opposite ends, so giving a 150mm x 160mm closed channel. Then, on the ends of the case, I'd run 120mm intake and an exhaust fans.
Sound adequate? Should be able to give 2kW with plenty of headroom.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.