I'm interested in a Leach Low-TIM amp with less power/lower voltage rails. WML recommends minimum +/-50VDC rails and the Low-TIM design is +/-58VDC (rated 120W/8 ohms). I'm targeting around 35W at 8 ohms. Just tired of the chip-amp (quasi-complimentary) sound proliferating in my TV, PC, portables etc. and thought of a decent say 35W solid-state amp. Apologies to the quasi crowd here. I built my double-barreled/Leach super-amp almost 30 years ago and love it.
The main changes I see are re-biasing the input diff-amp cascode and one or two pair of output transistors. TO-220 package looked good (compact) but On-Semi doesn't have much for TO-220 audio power transistors (which are basically the Low-TIM driver transistors MJE15030/MJE15031 $0.80). The NJW0281/NJW0302 (lower power 150W versions $1.96 of the NJW3281/NJW1302 200W are $2.38 parts at Newark. My point is you don't save anything other than space, so maybe leave the TO-264's alone? Has anyone scaled down this design?
The main changes I see are re-biasing the input diff-amp cascode and one or two pair of output transistors. TO-220 package looked good (compact) but On-Semi doesn't have much for TO-220 audio power transistors (which are basically the Low-TIM driver transistors MJE15030/MJE15031 $0.80). The NJW0281/NJW0302 (lower power 150W versions $1.96 of the NJW3281/NJW1302 200W are $2.38 parts at Newark. My point is you don't save anything other than space, so maybe leave the TO-264's alone? Has anyone scaled down this design?
In the beginning, like 30 years ago... I built three Low TIM Amps, one that was about 75 wpc and two more with 8 large heat sinks each at about 30 wpc. Both worked fine, but I had to crank the bias up on the lower power unit and blow a big fan on it, guess it ran about 10 wpc into Class A. I forget what happened to one of them, but one is deep in my closet.
Downscaling ususually does not save much money or space. IMHO it's most usefull if you want to cutdown the heat & power of a ClassA-amp or taylor it to your needs (high current/low rails for low impedances). For everything else you save maybe some outputs and that's it. But that's up to you 
Most well designed amps do not rely on the rails to define working points, in other words don't need any readjustments for lower or a bit higher rails. On a quick glance this seems also to be the case for the frontend of the lowTIM, but please check.
Have fun, Hannes
Most well designed amps do not rely on the rails to define working points, in other words don't need any readjustments for lower or a bit higher rails. On a quick glance this seems also to be the case for the frontend of the lowTIM, but please check.
Have fun, Hannes
I'll take it that no one has downsized a Leach Low-TIM amp. The Wall Street mantra, more power is better, lol. For some applications, I just don't need a lot of power and I thought noobs might like something less explosive.
I figure 60%-70% of the costs in a power-amp project are the power transformer, filter caps, heatsinks, chassis and downscaling definitely saves $$$ there. But on the amp PCB's, you wouldn't save more than a few bucks.
The Low-TIM input cascode has 40V (zener) Q-point, so rails less than +/-50V need changes to the input diff-amp and driver stage bias which is some design work. I'll try it on my next build and see how it turns out.
I figure 60%-70% of the costs in a power-amp project are the power transformer, filter caps, heatsinks, chassis and downscaling definitely saves $$$ there. But on the amp PCB's, you wouldn't save more than a few bucks.
The Low-TIM input cascode has 40V (zener) Q-point, so rails less than +/-50V need changes to the input diff-amp and driver stage bias which is some design work. I'll try it on my next build and see how it turns out.
The standard LO TIM is designed for ~+-60Vdc.
Reducing to +-50Vdc will require the resistors feeding the Zeners to be resized.
35W into 8r0 will work with 1pair of To247, To264, To3 or two pair To220.
35W is just 23.76Vpk into 8r0. You would need ~+-27Vdc or ~20+20Vac transformer. This will require the Zeners to be changed from 40V to ~20V and then the resistors feeding the Zeners must be resized and the resistors feeding the Tails must also be resized.
There are many much simpler and cheaper discrete designs that can put out between 35W and 100W that can compete with the Leach if they are competently built.
I run a pair of 60W chipamps. They cannot play loud. They become aggressive at much lower SPLs than a discrete amp delivering 100W. I normally listen @ <1W and it's the +20dB peaks that cannot pass through the amplifier without clipping that causes the sound quality problem.
Reducing to +-50Vdc will require the resistors feeding the Zeners to be resized.
35W into 8r0 will work with 1pair of To247, To264, To3 or two pair To220.
35W is just 23.76Vpk into 8r0. You would need ~+-27Vdc or ~20+20Vac transformer. This will require the Zeners to be changed from 40V to ~20V and then the resistors feeding the Zeners must be resized and the resistors feeding the Tails must also be resized.
There are many much simpler and cheaper discrete designs that can put out between 35W and 100W that can compete with the Leach if they are competently built.
I run a pair of 60W chipamps. They cannot play loud. They become aggressive at much lower SPLs than a discrete amp delivering 100W. I normally listen @ <1W and it's the +20dB peaks that cannot pass through the amplifier without clipping that causes the sound quality problem.
Last edited:
AndrewT, can you mention the alternate designs? Bipolar class AB 35W-100W ballpark would be okay. I'll take a looksee.
Over the years I've built a handful of discrete power amps and most were disappointing in terms of stability and fidelity, they ended up in the junk pile
Although the Leach design uses many transistors (over two dozen), perhaps twice as many as other designs, I don't worry about those costs.
I have the same experience with chip-amps, they "blare" at higher SPLs. Massive filter caps (>10kuF) and oversized power transformer help. I feel some of the chip-amp sound problem is due to the absence of PNP output transistors. That is a semi manufacturing limitation, something you cannot mix on a chip and why a discrete design can do better.
The Leach Amp is probably the all time #1 great DIY Amp - Kudos to Prof Leach - and will live on for another 30 years but probably morph into using more modern semiconductors. Its been going for over 30 years and there are a few things that would be done differently today, but it is a fine amp. Period.
I think AndrewT mentions some of the things that need attention, but I'll re-summarize here again:-
Use 1 single pair of TO-247/TO-264 Devices. Since the Leach Amp used fairly slow output devices, you can use 1 pair of MJL21194/21195. Because these are high voltage devices that you will be running on much lower rails, you get the benefit of reduced 2nd breakdown possibility.
Transformer - 20-0-20. Use 100VA component - should make for a very compact little 35W amp that can drive most domestic speaker loads.
Resize the input current source circuit. On 27-30VDC rails, you will need to use 18V-20V zener references to ensure under absolute worst case supply rail conditions you continue to get good current flow through them. Simply select the tail resistor (this is the resistor feeding each of the diff-amps) for the same current value as in the standard Leach Amp. Easiest way is to use only 1 1N5250 (20V 0.5W zener) per diff amp and then to put 2 12K resistors in parallel to get your 3mA tail current. A mismatch in the tail currents of 5% between the two diff amps only leads to a very slight increase in distortion (few PPM), so critical matching is not necessary.
Change the resistor that feeds the 1N5250 from the associated supply rail to 680Ohms (it is 2.2K in the standard Leach Amp). This gives a Zener current of 10mA - only slightly higher than the original 9mA.
For the small signal devices, you can go for BC557/556 across the board, but keep the VAS and the drivers the same.
I doubt there will be any compensation issues or stability issues.
Good luck
I think AndrewT mentions some of the things that need attention, but I'll re-summarize here again:-
Use 1 single pair of TO-247/TO-264 Devices. Since the Leach Amp used fairly slow output devices, you can use 1 pair of MJL21194/21195. Because these are high voltage devices that you will be running on much lower rails, you get the benefit of reduced 2nd breakdown possibility.
Transformer - 20-0-20. Use 100VA component - should make for a very compact little 35W amp that can drive most domestic speaker loads.
Resize the input current source circuit. On 27-30VDC rails, you will need to use 18V-20V zener references to ensure under absolute worst case supply rail conditions you continue to get good current flow through them. Simply select the tail resistor (this is the resistor feeding each of the diff-amps) for the same current value as in the standard Leach Amp. Easiest way is to use only 1 1N5250 (20V 0.5W zener) per diff amp and then to put 2 12K resistors in parallel to get your 3mA tail current. A mismatch in the tail currents of 5% between the two diff amps only leads to a very slight increase in distortion (few PPM), so critical matching is not necessary.
Change the resistor that feeds the 1N5250 from the associated supply rail to 680Ohms (it is 2.2K in the standard Leach Amp). This gives a Zener current of 10mA - only slightly higher than the original 9mA.
For the small signal devices, you can go for BC557/556 across the board, but keep the VAS and the drivers the same.
I doubt there will be any compensation issues or stability issues.
Good luck
Bonsai, thanks for your help! I'll start with 20V zeners, 1/2 the tail bias resistors and aim for around +/-28VDC rails.
I have to re-think if 4ohm loads would be alright with one pair of outputs. I found this on WML's website FAQ:
"Removing 1/2 the output transistors will not reduce the output power. It reduces the maximum output current by a factor of 1/2. For an 8 ohm load, the power output would not change. For lower load impedances, the protection circuits would kick in at a lower output current than with all 4 output transistors. If you want to operate the amplifier with 1/2 the output transistors, Q18 and Q19 are the ones to omit. You can also omit R37, R38, R41, R42, R45, and R46. These changes will reduce the maximum output current by a factor of 1/2. If you wish to decrease the output power, you must reduce the rail voltages. I do not recommend reducing them to less than +50 V and -50 V."
I have to re-think if 4ohm loads would be alright with one pair of outputs. I found this on WML's website FAQ:
"Removing 1/2 the output transistors will not reduce the output power. It reduces the maximum output current by a factor of 1/2. For an 8 ohm load, the power output would not change. For lower load impedances, the protection circuits would kick in at a lower output current than with all 4 output transistors. If you want to operate the amplifier with 1/2 the output transistors, Q18 and Q19 are the ones to omit. You can also omit R37, R38, R41, R42, R45, and R46. These changes will reduce the maximum output current by a factor of 1/2. If you wish to decrease the output power, you must reduce the rail voltages. I do not recommend reducing them to less than +50 V and -50 V."
I think the Leach Amp is the #1 DIY power amp. When the Audio Magazine article came out in 1980, it was a landmark. I think it should have been published in the AES Journal. DIY'ers were still burning after the SWTPC Tigersaurus disaster.
It's 30 years later and still worth all the effort building one. I'm going to start a thread on Leach Amp transistor alternates/substitutes. The TO-39 package is fading and I have been reading Doug Self's book on power amp design which has good info on the transistor requirements. I was wondering if the changes over the years are the best.
It's 30 years later and still worth all the effort building one. I'm going to start a thread on Leach Amp transistor alternates/substitutes. The TO-39 package is fading and I have been reading Doug Self's book on power amp design which has good info on the transistor requirements. I was wondering if the changes over the years are the best.
For what it s worth, I have completed a design - with full documentation - on a fully compliementary amp that borrows heavily from Leach, but with a more modern slant and components. PCB's are laid out etc, but I have not had the time to order or build them yet - just too busy at work and its set to be that way for at least another 2 months. I am a bit reluctant to publish just yet, having not built the board and tested it.
I have another big 250W amp that's been runnung for 3 years, so I am pretty confident my latest effort is good to go.
I have another big 250W amp that's been runnung for 3 years, so I am pretty confident my latest effort is good to go.
Last edited:
Bonsai, thanks for your help! I'll start with 20V zeners, 1/2 the tail bias resistors and aim for around +/-28VDC rails.
I have to re-think if 4ohm loads would be alright with one pair of outputs. I found this on WML's website FAQ:
"Removing 1/2 the output transistors will not reduce the output power. It reduces the maximum output current by a factor of 1/2. For an 8 ohm load, the power output would not change. For lower load impedances, the protection circuits would kick in at a lower output current than with all 4 output transistors. If you want to operate the amplifier with 1/2 the output transistors, Q18 and Q19 are the ones to omit. You can also omit R37, R38, R41, R42, R45, and R46. These changes will reduce the maximum output current by a factor of 1/2. If you wish to decrease the output power, you must reduce the rail voltages. I do not recommend reducing them to less than +50 V and -50 V."
I think the reason Prof. Leach does not recommned going below +-50 is that he is not proposing above to recalculate the LTP tail current setting resistors and the feed to the zener regulators. If you are preapred to do that, then you can reduce the rails to th elevel proposed by AndrewT plus some margin. Otherwise, I'd go with Prof Leach's recomendations on what to omit.
I have to re-think if 4ohm loads would be alright with one pair of outputs. I found this on WML's website FAQ:
"Removing 1/2 the output transistors will not reduce the output power. It reduces the maximum output current by a factor of 1/2. For an 8 ohm load, the power output would not change. For lower load impedances, the protection circuits would kick in at a lower output current than with all 4 output transistors. If you want to operate the amplifier with 1/2 the output transistors, Q18 and Q19 are the ones to omit. You can also omit R37, R38, R41, R42, R45, and R46. These changes will reduce the maximum output current by a factor of 1/2. If you wish to decrease the output power, you must reduce the rail voltages. I do not recommend reducing them to less than +50 V and -50 V."
I think the reason Prof. Leach does not recommned going below +-50 is that he is not proposing above to recalculate the LTP tail current setting resistors and the feed to the zener regulators. If you are preapred to do that, then you can reduce the rails to th elevel proposed by AndrewT plus some margin. Otherwise, I'd go with Prof Leach's recomendations on what to omit.
In my view an amp is designed based on current through every component. The exception is Vre at the output.
On this basis, currents in the leach must be found/calculated/measured.
These operating currents are then inserted into the low voltage rail version of the amplifier and all the resistances calculated to allow those "Leach" currents to flow.
Many of the design decisions made in an amplifier design are based on compromises that involve power dissipation and current optimised semi conductor parameters. That's the bit that can seriously affect final amplifier performance. It's current that matters.
This is quite different from our usual way of looking at amps that simply copy a working design. Here we look at and compare voltages. That's easy to do. We do not need to know the actual current because if the voltage is the same and the resistor is the same value then automatically the current must also be the same operating current.
As I said, "my opinion".
On this basis, currents in the leach must be found/calculated/measured.
These operating currents are then inserted into the low voltage rail version of the amplifier and all the resistances calculated to allow those "Leach" currents to flow.
Many of the design decisions made in an amplifier design are based on compromises that involve power dissipation and current optimised semi conductor parameters. That's the bit that can seriously affect final amplifier performance. It's current that matters.
This is quite different from our usual way of looking at amps that simply copy a working design. Here we look at and compare voltages. That's easy to do. We do not need to know the actual current because if the voltage is the same and the resistor is the same value then automatically the current must also be the same operating current.
As I said, "my opinion".
my first super leach amp was built using 2sd424/2sb554 output trannies on +/-70 volt rails....that was in 1984...
i have built a super leach amp using 2n3055/mj2955 on +/-55volts rails, i have tried different types of transistors....small signal and power....
it is the topolgy that has a lot to do with the sound imho....
i have built a super leach amp using 2n3055/mj2955 on +/-55volts rails, i have tried different types of transistors....small signal and power....
it is the topolgy that has a lot to do with the sound imho....
In resistively loaded LTP designs like the Leach, as long as the LTP current is held constant, the VAS standing current, pre-driver and driver standing currents are all fixed, and accurately.
Jens Rasmussen created a few Leach Amp clone boards that use modern devices. Component values carried over from the original, including compensation caps. As Tony pointed out, the circuit seems to work well with a variety of transistors. Several rounds of group buys didn't uncover any stability issues. I don't remember whether there was any effort to determine optimum values of compensation caps.
One thread is here http://www.diyaudio.com/forums/solid-state/52459-smaller-leach-amp-v1.html The original premise was to create a four plastic output device version, with lower output power than Jens' original 10 device clone.
One thread is here http://www.diyaudio.com/forums/solid-state/52459-smaller-leach-amp-v1.html The original premise was to create a four plastic output device version, with lower output power than Jens' original 10 device clone.
I missed the "Smaller Leach Amp V1" thread, but took the time to read the 154 pages/1,534 posts
- definitely a wealth of information. The only thing smaller about it is [physically] the move to plastic output transistors and some filter caps integrated on the one pcb. Instead of building a Double Barreled/Super-amp, they added a bunch more output transistors (6,10,12) looks like to increase the Low-TIM design to near the super-amp power levels.Just to summarize things, looking at running a Leach Low-TIM V4.5 at 1/2 the usual voltage, around +/- 28-30VDC rails, 4 ohm capable. People measured Vout max. (peak) = Rails - 6V. So 28-30V rails might result in 30-35W at 8 ohms, or 60-72W at 4 ohms. Changes so far:
The LTP tail resistors R15, R16 from 12K to 6.2K (WML's design notes: The cascode diff-amp stage has tail bias current of 3.25mA each).
Changing the input cascode bias from 40V to 20V zener. Lowering R13, R14 from 2k2 to 1-1.2kohm (my calcs), Bonsai mentioned 680 ohm. (WML's design notes: Each zener diode is biased at a current of about 3.3 mA). The VAS stage bias current is 4.2 mA and should adjust there, although it could be increased as Pdiss is much lower on Q12, Q13.
The bear is the protection circuit, a bit complex. I don't do a lot of PSPICE modeling. But with this much lower rail voltage, the SOA is less of a concern. Example MJW21192 125W/8A parts at 30V Vce, around 6A max./250ms (keeping 4 output transistors @50% SOA), which should conquer the rail fuses The MJL3281A 200W/15A is a whopping $0.25 more and 10A max./250ms, 7A continuous at 30V ea.
The V point (R21, R22) could come down, keeping the R21+R23 and R22+R24 totals the same. The I point leave the same, my guess it starts at 6.4A (R37, R39). The IV slope compensation (R30,R31), I'm not sure about. Part of me wants to omit the protection circuit.
The LTP tail resistors R15, R16 from 12K to 6.2K (WML's design notes: The cascode diff-amp stage has tail bias current of 3.25mA each).
Changing the input cascode bias from 40V to 20V zener. Lowering R13, R14 from 2k2 to 1-1.2kohm (my calcs), Bonsai mentioned 680 ohm. (WML's design notes: Each zener diode is biased at a current of about 3.3 mA). The VAS stage bias current is 4.2 mA and should adjust there, although it could be increased as Pdiss is much lower on Q12, Q13.
The bear is the protection circuit, a bit complex. I don't do a lot of PSPICE modeling. But with this much lower rail voltage, the SOA is less of a concern. Example MJW21192 125W/8A parts at 30V Vce, around 6A max./250ms (keeping 4 output transistors @50% SOA), which should conquer the rail fuses
The MJL3281A 200W/15A is a whopping $0.25 more and 10A max./250ms, 7A continuous at 30V ea.The V point (R21, R22) could come down, keeping the R21+R23 and R22+R24 totals the same. The I point leave the same, my guess it starts at 6.4A (R37, R39). The IV slope compensation (R30,R31), I'm not sure about. Part of me wants to omit the protection circuit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.