Other threads are talking about this amp, but they refer to older and multiple versions. Confusing.
In my opinion, this amp is so good that it deserves a thread alone.
I bought the DIY version on Aliexpress and replaced all components but the transistors, with good quality ones. I also discovered they are maintaining this 9.2 version, the PCB is updated compared to some other PCBs I saw some years ago.
This amp is based on D.Self publications, and with some super easy mods, it becomes a very good HiFi one.
Good points are the CFPs in the front end, and the triplets in the back end. Both refinements are absent in the newer V10 version.
Here is the schematic with the mods. The small resistors I used are all Vishay MBB/SMA 207 professional 1% 0.6W.
I love its sound, very precise, and never harsh. I'm using it all day long, and I compare its sound with the Circlotron's. Very close!
Cheers!
In my opinion, this amp is so good that it deserves a thread alone.
I bought the DIY version on Aliexpress and replaced all components but the transistors, with good quality ones. I also discovered they are maintaining this 9.2 version, the PCB is updated compared to some other PCBs I saw some years ago.
This amp is based on D.Self publications, and with some super easy mods, it becomes a very good HiFi one.
Good points are the CFPs in the front end, and the triplets in the back end. Both refinements are absent in the newer V10 version.
Here is the schematic with the mods. The small resistors I used are all Vishay MBB/SMA 207 professional 1% 0.6W.
I love its sound, very precise, and never harsh. I'm using it all day long, and I compare its sound with the Circlotron's. Very close!
Cheers!
Attachments
Are 220 Ohm and 470 Ohm marked in red non-existent before the mod? If so, I have no confidence if the board is a good design at the first place.
Me too. I completely agree about its sound quality. I have an amplifier using the L20 V9.2 boards running with 40V rails. I find the sound is utterly transparent and pure. It has no sound of its own as one CD sound can be totally different to another one. It really is the "straight wire with gain". The only thing I changed is the input capacitor for a large value film capacitor. I am always amazed at how cold it runs too. I do have large heatsinks on the boards but even driving 4 Ohms speakers at high volume they barely warm up, however there is no trace of crossover distortion sound, just purity at the very lowest levels that ensures the quietest ambience and acoustic is fully revealed. Years ago, it was said that a good amp should have a "black background sound". These amplifiers prove that is wrong. The acoustic is fully reproduced until the most minute sound is over (if ever) thus the background never becomes "black". You are "in the room" until the track finishes. I'm sure anyone hearing these amps would be impressed. PS: LJM himself on this site, said he thought it was his best sounding amp! PPS: Due to lowish quiescent power usage, I also have my amp running all day for quality background music. If it was a class A amp I would only use it for serious listening due to the high cost of electricity nowadays, A real bonus I think.
Last edited:
I'm about to build another one with the latest "Panda" boards. For anyone who does not know these boards there are two other sets of O/P trannies on the reverse side, thus making a total of 8 for each board
Attachments
Last edited:
8 OP total or 8 transistors total a channel?
I will build the real thing using a 140V total PSU, expecting an output power of about 200W/8Ohms. I know this way there is a transistor becoming hot, but I don't remember which one, perhaps T219, I'll find out.
I ordered the soft start and the speakers protection boards, to verify if I can fit 2 x 500VA toroids and all the rest in a 2U cabinet 360 x 300mm internal space.
I will build the real thing using a 140V total PSU, expecting an output power of about 200W/8Ohms. I know this way there is a transistor becoming hot, but I don't remember which one, perhaps T219, I'll find out.
I ordered the soft start and the speakers protection boards, to verify if I can fit 2 x 500VA toroids and all the rest in a 2U cabinet 360 x 300mm internal space.
The devil makes pans, not lids! A good design left incomplete with no apparent reason! All red components are added by me, not present in the original design, as they were supposed to be.
As said, I like this amplifier a great deal... However, maybe there is something even better to behold with your mods? Would you be so kind as to explain them on here one by one, please? The most important ones it seems are the ones shown in red. Why are they necessary and how do they improve the performance? I'm not concerned about the O/P coil, just the R251-R252 and C251? What does the extra T271 do also? Many thanks...PS: the latest V9.2 board uses 0.1 Ohm emitter resistors on the O/Ps the older version used 0.22 Ohms
Last edited:
Let's start from the blue mods:
C211 is a polypropylene capacitor replacing the original electrolytic one, which is not polarized by any DC; you cannot use an electrolytic capacitor when there is no DC at its ends; a very bad solution.
R213 replaces the original 10KΩ resistor, making the L-pad R212/R213 almost negligible, also raising the input impedance.
C216, R224, and R233 are the feedback circuitry scaled of a 5-factor (more or less) which enables a bipolar C216 to be used (a 1mF BP capacitor would have been too big to fit the PCB); C216 must be a bipolar one because there is no DC at its ends (do not consider the offset, which is very low), so the signal swings from positive to negative values and vice versa.
C219 should be a bipolar one because the DC at its ends is negligible.
R242 and C220 are the Zobel network: C220 can be crossed by a high current, so it has to be able to handle it, the original C220 is too small for this purpose.
Red mods:
R251 prevents T222 and T223 from turning completely off; the prime way to accomplish this sees R251's low end connected to the negative rail, and another R251 connected between R223's emitter and the positive rail, resulting in a Class-A operativity; but the 2 resistor would dissipate a lot of energy; this simplified solution is equivalent to the prime and only dissipates very low energy (D.Self "The blameless amplifier").
R252 has the same duty as R251 but is applied to T224 and T225.
C251: since the OPa are going to switch off during the opposite semi-cycle, they should do it fast, and this is the duty assigned to this capacitor (D.Self "The blameless amplifier").
L251 and R253: R253 is inside r251; after winding L251 it is better to put it inside a heat shrinking hose; for a complete understanding of L251 it is better to read Bob Cordell's "The Series L-R Network"; after this you perhaps will ask me why I used a 2.3µH value instead of 1µH one: the reason is that I cannot rely on a low open loop output impedance at high frequencies of this amp.
Green mod:
T271 is a beta enhancer. This amp doesn't strictly need this mod, because the output stage is made out of a triplet, which has a high current gain; I didn't try it because it is hard to fit it in the PCB, but it is reported on D.Self "the blameless amplifier" being the output stage a Darlington, not a Triplet. Try it at your risk!
DO NOT PROCEED TO ANY MOD that implies desoldering of components from the PCB if you don't have a reliable desoldering pump, the PCB is very delicate, and attempts to fit a lead in a closed hole in the PCB will damage it for sure.
Thank you for updating me. I like the 0.1Ω emitter resistors, lowering needed bias, and the cross-over distortion. Unfortunately, they also lower the local feedback of the OP transistors, exposing them to a possible thermal runaway when you use high voltage to feed this amp. However I got the DIY kit about 10 days ago, and the Res are 0.22Ω.
C211 is a polypropylene capacitor replacing the original electrolytic one, which is not polarized by any DC; you cannot use an electrolytic capacitor when there is no DC at its ends; a very bad solution.
R213 replaces the original 10KΩ resistor, making the L-pad R212/R213 almost negligible, also raising the input impedance.
C216, R224, and R233 are the feedback circuitry scaled of a 5-factor (more or less) which enables a bipolar C216 to be used (a 1mF BP capacitor would have been too big to fit the PCB); C216 must be a bipolar one because there is no DC at its ends (do not consider the offset, which is very low), so the signal swings from positive to negative values and vice versa.
C219 should be a bipolar one because the DC at its ends is negligible.
R242 and C220 are the Zobel network: C220 can be crossed by a high current, so it has to be able to handle it, the original C220 is too small for this purpose.
Red mods:
R251 prevents T222 and T223 from turning completely off; the prime way to accomplish this sees R251's low end connected to the negative rail, and another R251 connected between R223's emitter and the positive rail, resulting in a Class-A operativity; but the 2 resistor would dissipate a lot of energy; this simplified solution is equivalent to the prime and only dissipates very low energy (D.Self "The blameless amplifier").
R252 has the same duty as R251 but is applied to T224 and T225.
C251: since the OPa are going to switch off during the opposite semi-cycle, they should do it fast, and this is the duty assigned to this capacitor (D.Self "The blameless amplifier").
L251 and R253: R253 is inside r251; after winding L251 it is better to put it inside a heat shrinking hose; for a complete understanding of L251 it is better to read Bob Cordell's "The Series L-R Network"; after this you perhaps will ask me why I used a 2.3µH value instead of 1µH one: the reason is that I cannot rely on a low open loop output impedance at high frequencies of this amp.
Green mod:
T271 is a beta enhancer. This amp doesn't strictly need this mod, because the output stage is made out of a triplet, which has a high current gain; I didn't try it because it is hard to fit it in the PCB, but it is reported on D.Self "the blameless amplifier" being the output stage a Darlington, not a Triplet. Try it at your risk!
DO NOT PROCEED TO ANY MOD that implies desoldering of components from the PCB if you don't have a reliable desoldering pump, the PCB is very delicate, and attempts to fit a lead in a closed hole in the PCB will damage it for sure.
Thank you for updating me. I like the 0.1Ω emitter resistors, lowering needed bias, and the cross-over distortion. Unfortunately, they also lower the local feedback of the OP transistors, exposing them to a possible thermal runaway when you use high voltage to feed this amp. However I got the DIY kit about 10 days ago, and the Res are 0.22Ω.
Very many thanks marigno. I've spent a few hours going over everything you say above and checking back on the Douglas Self Blameless design too. Very interesting it all is. One thing for C216, I checked the cap size of the Muse ES BP . The 1000uF 6.3V can size is 12.5mm x 25mm so you might be able to squeeze it in and then not have to change R233 and R244. I did check the emitter resistors, and they are 0.1 Ohms and the board is new and carries the "Panda" logo. As you say thermal runaway needs to be closely monitored with high rail voltages. I'm sure your mods will be of interest to other L20 V9.2 users. Final thoughts. I would love to hear what others say (good or bad) about this board, especially sound-wise.
If you mean LJM (its easy to get it the wrong way), then you might be lucky if you send a direct message to him see his details: https://www.diyaudio.com/community/members/ljm_ljm.54673/
Thank you, I was registered on the LJM site, but can't find it anymore.
I checked my new boards, and yes! they have 0.1Ω Res, I didn't pay attention to them. However the voltage does not exceed 110V, and the power supply is weak in current. I'm comparing the sound of this configuration with this one: https://www.diyaudio.com/community/...-a-hybrid-main-amplifier.396135/#post-7276068 and I must say L20's sound is close to the WHA-53's.
My new project, as I intend it, will have a 100V/500VA transformer per channel, making available a power of 310W on 140V (126V under load) once rectified. I will build it only if I can fit everything in a 2U cabinet (HiFi2000's Dissipante 300mm). I'll keep the 0.1Ω Res as an option to try.
I checked my new boards, and yes! they have 0.1Ω Res, I didn't pay attention to them. However the voltage does not exceed 110V, and the power supply is weak in current. I'm comparing the sound of this configuration with this one: https://www.diyaudio.com/community/...-a-hybrid-main-amplifier.396135/#post-7276068 and I must say L20's sound is close to the WHA-53's.
My new project, as I intend it, will have a 100V/500VA transformer per channel, making available a power of 310W on 140V (126V under load) once rectified. I will build it only if I can fit everything in a 2U cabinet (HiFi2000's Dissipante 300mm). I'll keep the 0.1Ω Res as an option to try.
Your article on the WHA-53 is superb, so high praise indeed for the L20. PS: I would like "Audio Science" to test one and give his rating... PS: Any hands-on experience with the L12-2 AMP? The basic design (Blameless) is similar but different O/P trannies. The L12 amp gets much more coverage and recommendations... I wonder why?
Does it matter what “he” says? It’s your ears that matter 🙂.
Don’t trust your ears. There is a thing called auditory psychology. Manufacturer can take advantage of this to make distorted sound appear more pleasing. Bose is a master of this.
Last edited:
I couldn’t disagree more.
But, I don’t want to derail this thread and won’t comment any further.
I apologize Marigno.
But, I don’t want to derail this thread and won’t comment any further.
I apologize Marigno.
Usually, I don't pay too much attention to designs with a global feedback and xVbe. I'm studying a balanced VAS that gives 60Vp at its output, to drive a Circlotron buffer. No conventional feedback, no xVbe. It uses Schade feedback. After putting again my hands on the L20, I must say I have to reconsider my interests including this type of amps. However, it is easier to obtain good results from a tube design.
Good point, I see this is the best practice. R213 can be 47KΩ, but I would change R212 to 1KΩ.
Hi Vunce, I have some news you could be interested in. I'll send you an email.
You don't have to apologize, you have the same opinion as mine on the sound, and this is the reason I limit my posting here, I like to design audio machines, the "education" is not my best. I am back to post on the L20 because it is not a design of mine and it is so good after all mods I couldn't stay still. It deserves a thread by itself. And I learned something, as usual, look at posts by @sonicles and post #14.