redjr, all I can say is that we have to accept a certain measure of uncertainty because even if we are blessed with the opportunity to borrow someone else´s diy amp it´s not that sure that if we build a similar amp the end result will be the same. I have spent numerous hours considering many alternatives for an amp to build as I need one more mono amp at least to finish what I now realize was a far to ambitious project. But if I got the money I would probably go for another one of a certain commercial amp I own.
Building amps are some sort of voyage of discovery where you often have too little information and if you are not prepered for some unexpected experiences you may reconsider this hobby.
It´s like the bank employee that answer the following question from an old couple: "we want ot buy shares but we are not prepered to take risks". The straight answer is; "then I strongly advise you not to buy shares."
I may have got it wrong but to me it seems that you asipire to a certain success that I might consider unachievable. But this may reflect my own frusration as well.
Building amps are some sort of voyage of discovery where you often have too little information and if you are not prepered for some unexpected experiences you may reconsider this hobby.
It´s like the bank employee that answer the following question from an old couple: "we want ot buy shares but we are not prepered to take risks". The straight answer is; "then I strongly advise you not to buy shares."
I may have got it wrong but to me it seems that you asipire to a certain success that I might consider unachievable. But this may reflect my own frusration as well.
Terry,
Aren't the KSA 100 Krell? I thought those amps were only commercial/retail and not available for DIY. Do they offer just the PCB? I was reading another thread (it's easy to get them confused at times.) where they were discussing the KSA 100 I believe, but didn't get the connection between retail Krell and the KSA 100. Are they clones of what's in a Krell? IIRC some young guy in the Netherlands was building one, and it was humongous and looked to weight about 100 pounds! Not that that would deter me
Hi Red,
Yes, the KSA is a Krell. They are older models that I suppose are past the end of the patent. There are a few threads on here about them. Class A, fairly easily available parts, nothing exotic. They sound very good. The biggest expense is the power supply and the heatsinks. There are PCB's available on ebay.
Look here.
Hey thanks Terry. I'll check out ebay. Are the PCBs genuine Krell discounted discontinued overstock, or Asian knock-off?
Weird, the link I gave didn't work. You will have to search for "
High Power Pure Class A amplifier PCB KSA100"
Certainly not Krell NOS. They are perhaps "Asian Knockoff". I don't know where they got the gerber files to make them. They are different than mine which where part of a group buy offered by Mark Gulbrandsen. I have not yet built mine as I wasn't sure I wanted to spend as much for the transformers and heatsinks as it would take to carry it off. Now holding them in my hand, I am getting the urge to do it. I'm tempted to strip the Aleph-X and use the case and heatsinks from that to built it.
High Power Pure Class A amplifier PCB KSA100"
Certainly not Krell NOS. They are perhaps "Asian Knockoff". I don't know where they got the gerber files to make them. They are different than mine which where part of a group buy offered by Mark Gulbrandsen. I have not yet built mine as I wasn't sure I wanted to spend as much for the transformers and heatsinks as it would take to carry it off. Now holding them in my hand, I am getting the urge to do it. I'm tempted to strip the Aleph-X and use the case and heatsinks from that to built it.
Terry,
For some odd reason ebay links are not working. Not sure if that's an ebay issue, or something has changed on the forum. My son informed me last night that you have to use something like tiny.url to shorten the address and then it seems to work. (He has all the answers!
) We tested it on a post and it does works. Go to www.goo.gl for more info.Thanks for the info on the Krell boards. I'll search on ebay and see what turns up.
Hello everybody!
Please excuse my absence. I left town and had no internet what so ever.
I looked over this thread again and again and again and again and again and came to the final conclusion. It's either the Blameless amp (trimordial amp?) or the Leach Superamp. Blameless is preferred tho' because of the topology and because of not using cascaded output devices.
What i'm looking for:
- +-80v operation
- 8ohm capable. 4 ohm wouldn't be bad but not necessary
- enough output devices so it's reliable but not too many as it would get much too expensive
- extremely good sound
- protections? can be added as a standalone pcb anyway (i was thinking of the PIC protection)
- LIMITER. I just can't stress this enough. A limiter is a must.
If someone, anyway could supply me with a schematic having the above mentioned characteristics, i'd be so happy i'd literally start sh****g rainbows! So please, if by any means, you come across a Blameless amp (preferred) or a Superamp (less preferred) schematic with those characteristics, i'm begging you, post it here or send me a pm.
All my regards,
Matei.
Please excuse my absence. I left town and had no internet what so ever.
I looked over this thread again and again and again and again and again and came to the final conclusion. It's either the Blameless amp (trimordial amp?) or the Leach Superamp. Blameless is preferred tho' because of the topology and because of not using cascaded output devices.
What i'm looking for:
- +-80v operation
- 8ohm capable. 4 ohm wouldn't be bad but not necessary
- enough output devices so it's reliable but not too many as it would get much too expensive
- extremely good sound
- protections? can be added as a standalone pcb anyway (i was thinking of the PIC protection)
- LIMITER. I just can't stress this enough. A limiter is a must.
If someone, anyway could supply me with a schematic having the above mentioned characteristics, i'd be so happy i'd literally start sh****g rainbows! So please, if by any means, you come across a Blameless amp (preferred) or a Superamp (less preferred) schematic with those characteristics, i'm begging you, post it here or send me a pm.
All my regards,
Matei.
Hi golden - Never got around to replying to this... I'm actually not after audio/sonic nirvana with my next amp build. My quest is simply to raise the bar - of both my project building skills - while at the same time hopefully getting a 'more than decent' top-notch power amp. I haven't tackled a discrete output amp yet, so that's what's drawing me to the Honey Badger. I have no allusions over the road-blocks I might confront, and I'm fairly confident that my precision approach at populating PCB's will yield something that will work correctly. I look at the total project build, not just the amp modules themselves. I'm equally as interested in a professional looking chassis with matching internal layout as well. I never rush these things, and probably go overboard with the detailed planning. Will I be able to measure the specs when it's all said and done? No. I'll use my ears for first pass testing.
So, for the moment, the HB gets the nod on my next project. The PCBs and parts have been ordered, so just waiting for everything to get here. Then it will all be on hold as I'm moving my household to another state. What fun.
Rick
Hi Guys
Brlmat, see post 92 for the parts to use in the Blameless for the voltage and power you want.
Less than four pairs of outputs and you won't have a reliable amp at 250W-8R and 500W-4R.
The HB has compromises and added setup complexity both of which detract from performance. To me, the deviations chosen make the HB not worth building.
The basic Blameless has little or no offset DC voltage at its output - especially if you add a 1k to ground from the output. ALL amps should have this resistor in place but none do except for my own. So, there is no need for an offset adjustment pot.
The HB has compromised current sources. The one that is recognisable is not truly 'active', rather it is buffered passive. The VAS works against a bootstrap instead of a true current source, so its performance and gain are reduced. Overall, THD will be much worse with the HB than with Self's recommended form, and slew rate will be nearly half what it should be.
It is human nature to take something great and want to change it.
For your app, the only changes to the Blameless required are to make the output stage a triple and use four pairs of outputs. Then, use the high-voltage parts listed previously so the circuit is compatible with your +/-80V rails. I can send you a schematic of this.
Have fun
Kevin O'Connor
Brlmat, see post 92 for the parts to use in the Blameless for the voltage and power you want.
Less than four pairs of outputs and you won't have a reliable amp at 250W-8R and 500W-4R.
The HB has compromises and added setup complexity both of which detract from performance. To me, the deviations chosen make the HB not worth building.
The basic Blameless has little or no offset DC voltage at its output - especially if you add a 1k to ground from the output. ALL amps should have this resistor in place but none do except for my own. So, there is no need for an offset adjustment pot.
The HB has compromised current sources. The one that is recognisable is not truly 'active', rather it is buffered passive. The VAS works against a bootstrap instead of a true current source, so its performance and gain are reduced. Overall, THD will be much worse with the HB than with Self's recommended form, and slew rate will be nearly half what it should be.
It is human nature to take something great and want to change it.
For your app, the only changes to the Blameless required are to make the output stage a triple and use four pairs of outputs. Then, use the high-voltage parts listed previously so the circuit is compatible with your +/-80V rails. I can send you a schematic of this.
Have fun
Kevin O'Connor
He means the after the VAS, there are 3 stages till output stage itself.
Basically OP stage
Driver
Pre-Driver
is your EF3 or triple.
And how long has the HB been around? Couple of years now. And this is the first I'm hearing of this? Are these 'compromises' you speak of ones that affect the sonic qualities of the amp or just there to make life miserable for us poor DIY souls?
I don't discount what you are saying. I can't. Unless all of this is your opinion, it seems it would have been recognized by other knowledgeable people - who know about these things and mentioned as either worthless, or at best unnecessary! And discovering all of this just after I purchased the boards and BOM. Urgh!Hi Guys
Personally I had not looked at the HB until just last week or so, as it was mentioned as a possible solution for Brlmat.
The changes made are indeed ones that reduce performance. As I said, slew rate is easily cut in half with one of the changes.
Most hobbyists try to reduce transistor count, because "Oh you don't really need that" and "I can simplify this..." When you do not know what you are looking at, it is easy to think that "such a tiny cap" or "such a small resistor" could easily be removed without impairing the circuit operation. Or that changing out an active current source with an entirely passive form requiring critical resistor values will somehow not change anything. Those little refinements are what makes the circuit practical with a less than perfect layout., and the not-so-little things are what makes the amp have great performance instead of simply being functional.
Self released his book in "editions", adding articles published in magazines that came out after the previous edition of the book. So, it is possible that when the HB project was started, some of the distortions over eight were not addressed. However, the main points above are all part of even the first edition of the book, so it would be impossible for someone to overlook them. Before being published in the book, all eight original distortions and the Blameless amplifier concept were published in a series of articles in Electronics World magazine, printed in the UK but distributed world wide. Again, no excuse not to know about since the HB is based on the Blameless.
Once you get your boards, Rick, you can easily make some simple mods to return the circuit to the intended or potential performance. You may not hear a difference but you have that opportunity.
I have not read the HB threads, so it would be rather surprising for no one to ever have suggested being truer to the Blameless might be a good thing.
Have fun
Kevin O'Connor
Personally I had not looked at the HB until just last week or so, as it was mentioned as a possible solution for Brlmat.
The changes made are indeed ones that reduce performance. As I said, slew rate is easily cut in half with one of the changes.
Most hobbyists try to reduce transistor count, because "Oh you don't really need that" and "I can simplify this..." When you do not know what you are looking at, it is easy to think that "such a tiny cap" or "such a small resistor" could easily be removed without impairing the circuit operation. Or that changing out an active current source with an entirely passive form requiring critical resistor values will somehow not change anything. Those little refinements are what makes the circuit practical with a less than perfect layout., and the not-so-little things are what makes the amp have great performance instead of simply being functional.
Self released his book in "editions", adding articles published in magazines that came out after the previous edition of the book. So, it is possible that when the HB project was started, some of the distortions over eight were not addressed. However, the main points above are all part of even the first edition of the book, so it would be impossible for someone to overlook them. Before being published in the book, all eight original distortions and the Blameless amplifier concept were published in a series of articles in Electronics World magazine, printed in the UK but distributed world wide. Again, no excuse not to know about since the HB is based on the Blameless.
Once you get your boards, Rick, you can easily make some simple mods to return the circuit to the intended or potential performance. You may not hear a difference but you have that opportunity.
I have not read the HB threads, so it would be rather surprising for no one to ever have suggested being truer to the Blameless might be a good thing.
Have fun
Kevin O'Connor
Hi Guys
Brlmat, you don't have an email, so I will post the schemo below.
Rick, you can modify your HB boards to provide the full performance potential of the Blameless. You may not hear a difference but you will have the opportunity.
I glanced at the HB threads a while back, saw they were just an impaired Blameless build and went away. It would be surprising if no one ever suggested that being truer to the original might be beneficial. The deviations DO impair performance as stated above - not opinion, fact. Slew rate will be much reduced and THD will be higher.
There is no need for the offset adjust pot as the Blameless has practically zero offset without using matched devices. My builds of a 250W amp back in the early 90s had offset too low for my meter to resolve, so less than 2mV. It's really not a concern.
The circuit posted here uses active current sources to assure high slew rate. It uses an output triple EF - a common way to do things but not my first choice. There is a simple protection circuit, and a provision for an anti-clip circuit to be added (will post that later).
Have fun
Kevin O'Connor
Brlmat, you don't have an email, so I will post the schemo below.
Rick, you can modify your HB boards to provide the full performance potential of the Blameless. You may not hear a difference but you will have the opportunity.
I glanced at the HB threads a while back, saw they were just an impaired Blameless build and went away. It would be surprising if no one ever suggested that being truer to the original might be beneficial. The deviations DO impair performance as stated above - not opinion, fact. Slew rate will be much reduced and THD will be higher.
There is no need for the offset adjust pot as the Blameless has practically zero offset without using matched devices. My builds of a 250W amp back in the early 90s had offset too low for my meter to resolve, so less than 2mV. It's really not a concern.
The circuit posted here uses active current sources to assure high slew rate. It uses an output triple EF - a common way to do things but not my first choice. There is a simple protection circuit, and a provision for an anti-clip circuit to be added (will post that later).
Have fun
Kevin O'Connor
Hi Guys
Had not intended to post two nearly identical posts, but something happened after I wrote the first one and I thought it had been lost.
Maybe some things are worth reading twice. Have quite a few books I've read many times...
R47: 10R-1W in parallel with choke. Can be #22 wrapped tight spaced around resistor body for full length, ends soldered to coil ends.
R48: 1k00
R49: 10R 1W-5W
C7: 100nF-250V
protection R values may need tweaking
R56 more like 1k
Have fun
Kevin O'Connor
Had not intended to post two nearly identical posts, but something happened after I wrote the first one and I thought it had been lost.
Maybe some things are worth reading twice. Have quite a few books I've read many times...
R47: 10R-1W in parallel with choke. Can be #22 wrapped tight spaced around resistor body for full length, ends soldered to coil ends.
R48: 1k00
R49: 10R 1W-5W
C7: 100nF-250V
protection R values may need tweaking
R56 more like 1k
Have fun
Kevin O'Connor
Last edited:
Thanks for the schematic! May i ask where you got it from?
I will edit it and add the correct values to some components.
Can i use 2SC5200 and 2SA1943 as output devices?
How does the "To anti-clip circuit" connection work?
Q1 and Q2 should be thermally coupled? Any other?
Q11, Q15 and Q16 should be on the same radiator as the output devices, right?
Q13 and Q14 whould have their own small rad?
R27 is the bias pot right?
R48, R49 and C7 should be on the terminal post, right? or it doesn't matter?
What's the value of C4? 2,2 milifahrad ?
D1 to D4 should be on the big radiator or free air?
Sorry for all those questions. I went through so many schematics i can't figure out all of it cause i have so many designs in my head right now.
Thanks again,
Matei
I will edit it and add the correct values to some components.
Can i use 2SC5200 and 2SA1943 as output devices?
How does the "To anti-clip circuit" connection work?
Q1 and Q2 should be thermally coupled? Any other?
Q11, Q15 and Q16 should be on the same radiator as the output devices, right?
Q13 and Q14 whould have their own small rad?
R27 is the bias pot right?
R48, R49 and C7 should be on the terminal post, right? or it doesn't matter?
What's the value of C4? 2,2 milifahrad ?
D1 to D4 should be on the big radiator or free air?
Sorry for all those questions. I went through so many schematics i can't figure out all of it cause i have so many designs in my head right now.
Thanks again,
Matei
Hi Guys
Yes, that is the Blameless with performance enhancements suggested by its originator useful for the high voltages and power required here. Most of the designs in Self's book are for 20-50W, using a single pair of outputs. He shows how adding more outputs and going to the triple reduces distortion and allows low-z load handling. You need all of this anyway to handle high power.
Q7,8,13,14 do not require heat sinks, but are large packages to be reliable in their circuit locations. A TO-92 might die from the heat.
Q15,16 need heat sinks but should not be bolted to the main heat sink.
Q11 should be bolted to the main heat sink for thermal feedback.
1/4W Rs can be used almost everywhere.
Elevate 1W+ Rs 6mm (0.25") above the board.
Matei, zoom in on the upper and lower right hand corners and you see +80 and -80 designated for the supply rails.
Terry, many alternate device types can be used in any position. The ones you list are fine.
Q27 cascodes the tail CS to eliminate hum break through.
Q28,29 cascode the input diff to reduce distortion in the diff by lowering its temperature, while also greatly improving common-mode rejection.
Using only keyboard symbols:
u = micro
m = milli
k = kilo
M = mega
K = Kelvin degrees and is right out
so:
220u is 220 microfarad - farad is "understood" as it is the unit of capacitance, and is sitting beside a cap symbol.
Similarly, power supply filter values referred to in a previous post as 10-15mF, are in millifarads - 10,000u to 15,000u.
For the following, Vq is the quiescent idle voltage measured across the output stage emitter resistors, from the E on an upper device to an E on a lower device. So, Vq spans two 100m Rs in series.
The only pot is the bias trim pot. Were you to set Vq=25mV, then idle per output device would be 125mA with the emitter resistors shown (100mR). But... this is about what you need in an EF. At 80V rails, each device dissipates 10W; 80W at idle for the whole amp.
Here is where Self and leach differ: Self does not state whether that 125mA can be spread over the whole output stage or if it is per device. Leach suggests it can be spread over the output stage. That would reduce the target Vq to about 6mV. Other evidence suggests that the individual currents should be higher. Alternatively, 100m can be changed to 220m setting Vq to 25mV gives 56mA per device. In general, you see 50-100mA in EF devices exhibiting low THD.
The output resistor, choke and zobel can be on the board. There is little power dissipated by them. You can place them off the board if you wish, but tying too much stuff to the speaker connectors makes changing that connector more difficult, if you have to later.
D1-4 just protect the feedback cap from excess DC voltage if things go awry.
The anti-clip circuit monitors the diff-amp base voltages as this is the most reliable place in the amp to detect distortions. That circuit is a quad opamp and either a jfet or a transconductance opamp - the latter is noisier than the jfet. The circuit dials down the input signal to the amp to prevent clipping. I'll post this later. However, adding the two Rs shown is the only provision required in the basic power amp circuit.
Overall, the circuit looks complex but it is just a modified Lin with modern updates to improve slew rate, reduce THD to almost unmeasurable values, and provide reliable performance with no critical parts. Nothing requires matching. You'll notice the absence of a "speed-up" cap across R36. That cap can cause cross-conduction problems in real life circuits.
Have fun
Kevin O'Connor
Yes, that is the Blameless with performance enhancements suggested by its originator useful for the high voltages and power required here. Most of the designs in Self's book are for 20-50W, using a single pair of outputs. He shows how adding more outputs and going to the triple reduces distortion and allows low-z load handling. You need all of this anyway to handle high power.
Q7,8,13,14 do not require heat sinks, but are large packages to be reliable in their circuit locations. A TO-92 might die from the heat.
Q15,16 need heat sinks but should not be bolted to the main heat sink.
Q11 should be bolted to the main heat sink for thermal feedback.
1/4W Rs can be used almost everywhere.
Elevate 1W+ Rs 6mm (0.25") above the board.
Matei, zoom in on the upper and lower right hand corners and you see +80 and -80 designated for the supply rails.
Terry, many alternate device types can be used in any position. The ones you list are fine.
Q27 cascodes the tail CS to eliminate hum break through.
Q28,29 cascode the input diff to reduce distortion in the diff by lowering its temperature, while also greatly improving common-mode rejection.
Using only keyboard symbols:
u = micro
m = milli
k = kilo
M = mega
K = Kelvin degrees and is right out
so:
220u is 220 microfarad - farad is "understood" as it is the unit of capacitance, and is sitting beside a cap symbol.
Similarly, power supply filter values referred to in a previous post as 10-15mF, are in millifarads - 10,000u to 15,000u.
For the following, Vq is the quiescent idle voltage measured across the output stage emitter resistors, from the E on an upper device to an E on a lower device. So, Vq spans two 100m Rs in series.
The only pot is the bias trim pot. Were you to set Vq=25mV, then idle per output device would be 125mA with the emitter resistors shown (100mR). But... this is about what you need in an EF. At 80V rails, each device dissipates 10W; 80W at idle for the whole amp.
Here is where Self and leach differ: Self does not state whether that 125mA can be spread over the whole output stage or if it is per device. Leach suggests it can be spread over the output stage. That would reduce the target Vq to about 6mV. Other evidence suggests that the individual currents should be higher. Alternatively, 100m can be changed to 220m setting Vq to 25mV gives 56mA per device. In general, you see 50-100mA in EF devices exhibiting low THD.
The output resistor, choke and zobel can be on the board. There is little power dissipated by them. You can place them off the board if you wish, but tying too much stuff to the speaker connectors makes changing that connector more difficult, if you have to later.
D1-4 just protect the feedback cap from excess DC voltage if things go awry.
The anti-clip circuit monitors the diff-amp base voltages as this is the most reliable place in the amp to detect distortions. That circuit is a quad opamp and either a jfet or a transconductance opamp - the latter is noisier than the jfet. The circuit dials down the input signal to the amp to prevent clipping. I'll post this later. However, adding the two Rs shown is the only provision required in the basic power amp circuit.
Overall, the circuit looks complex but it is just a modified Lin with modern updates to improve slew rate, reduce THD to almost unmeasurable values, and provide reliable performance with no critical parts. Nothing requires matching. You'll notice the absence of a "speed-up" cap across R36. That cap can cause cross-conduction problems in real life circuits.
Have fun
Kevin O'Connor
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.