"Where can you get KEC KTB817 KTD1047?"
Actually, I found them at UTSource.
Are the ones that ship with LJM's board fine, or should they be replaced as well?
Actually, I found them at UTSource.
Are the ones that ship with LJM's board fine, or should they be replaced as well?
I haven't observed any difference in performance. In fact, I've kept devices from the kit in one channel and ISC products in the other. Measurement results were idendical, both were good.
But if you have own devices and larger number of them, then you may match them and that makes real difference. Also, other transistors make difference. More tomorrow, now I go to bed.
Last edited:
I have purchased 50 each small transistors and paired them all, also I have replaced delivered 1659A/4370A transistors with paired 1930/5171. The delivered two 669 seemed suspiciously faraway from eachother in both hfe and ube, so I have replaced them with ISC 669, which isn't plastic so I had to isolate them from the heat sink. All in all: I have paired all transistors, used 0,1% resistors and the result is virtually zero DC drift and 5-6 mV vrms at the output with signsl absent. I have used 42 V capacitance multiplier and it is conceivable that I had practically zero ripple. Also, all even harmonics have vanished.
My reasoning was that maticulous selection and pairing of devices can't harm.
I was able to make comparison before-and-after, with kit components and those purchased from Reichelt. The difference was obvious, in favor of the former.
Please, note that I am not bashing the kit, quite the contrary. This amplifier is great but requires a bit of effort to extract the maximum from it. For the bagatelle price, this kit is a gift. It would be shame to require paired first quality hand-picked devices in that price class.
Replacement of devices with measured and paired can help to make this amplifier shine. Of course, add to that bias adjustment, a decent power supply, heat sinks and a chasis and you will end-up with several hunderd Euro expenditure but also you may become a proud owner of a great amplifier.
My reasoning was that maticulous selection and pairing of devices can't harm.
I was able to make comparison before-and-after, with kit components and those purchased from Reichelt. The difference was obvious, in favor of the former.
Please, note that I am not bashing the kit, quite the contrary. This amplifier is great but requires a bit of effort to extract the maximum from it. For the bagatelle price, this kit is a gift. It would be shame to require paired first quality hand-picked devices in that price class.
Replacement of devices with measured and paired can help to make this amplifier shine. Of course, add to that bias adjustment, a decent power supply, heat sinks and a chasis and you will end-up with several hunderd Euro expenditure but also you may become a proud owner of a great amplifier.
Last edited:
So, let's start now!
3. Adjust the trimmer to exactly 1KOhm. This is very important - do not be sloppy here: exactly one kiloOhm.
10. a With oscilloscope
With the use of oscilloscope adjustment is much easier because you can see when distortion of the response to square wave ceases to exist. First stretch the rise line of the square wave to the width of almost entire screen. Initially there is a bulge, concavity which gradually fades with the increase of bias and then vanishes and turns the response into a smooth, convex line.
In my particular case I have increased voltage from 1.2 mV to 15mV across the R050 resistor.
That's all folks.
Enjoy the weekend!
So, in order to best use the boards they way they ship without completely rebuilding them, adjusting with the trimmer resistor is the best way to go?
Would this be a good choice for trimmer resistor?
Blocked
I have an oscilloscope (but don't have much experience using it), what are the instructions on how to use it to best set the trimmer? Where to connect it?
Thank you,
David.
Hi David,
Excuse me for delayed response. Meanwhile, I have focused my attention to finalizing A60 Accuphase clone.
That potentiometter is fine. Simply bend the wiper pin as close to the root and solder it to one of the remaining pins with a small drop of solder. Then adjust the potentiometer to 1KOhm. Solder it in place of the removed fixed value resistor. Measure voltage (mV) between the "OUT" connector and a middle pin of any of power transistors. Slowly adjust the voltage to 20 to 30 mV. Oscilloscope isn't mandatory but on the oscilloscope you can visualise fading distortion.
I am puzzled with designer decision to make this otherwise fine amplifier board as bare-bones, even without possibility to adjust it. Adequate amount of bias is necessary to achieve minimal distortion whilst increasing its stability. At low bias this board persistently oscillates at ridiculous 53MHz. It is true that this amplifier sounds good enough at low bias but there is a good chance to make it sound even better and run stable.
On oscilloscope you may place the rising edge of square wave response at the center of the screen and then zoom in untill yo see a straight line turning into an arc. With low bias there is concavity, a bulge. The increase of bias turns this irregular shape into a smooth, convex curve.
I hope this helps.
I wish you a pleasant Sunday afternoon.
Last edited:
L20.5 resistor change again
Hi Berlusconi,
Just letting you know I have reverted back to the 910 ohm resistors on the L20.5 boards not because of any problems but when I put the 953 or 976 ohm resistors on the L20.5 I felt I lost too much of their sound quality.
As you suggested I used a couple of pretty big heatsinks and they barely heat up at all.
I also found the millivolts had dropped a little from where they were when I first used the 910 ohm resistors so I am going to see how well the boards last in their current specification.
Many thanks again for all your help.
Bill :
Hi Berlusconi,
Just letting you know I have reverted back to the 910 ohm resistors on the L20.5 boards not because of any problems but when I put the 953 or 976 ohm resistors on the L20.5 I felt I lost too much of their sound quality.
As you suggested I used a couple of pretty big heatsinks and they barely heat up at all.
I also found the millivolts had dropped a little from where they were when I first used the 910 ohm resistors so I am going to see how well the boards last in their current specification.
Many thanks again for all your help.
Bill :
We have spectrum analysers at work and one of them works all the way down to the audio band. I am not sure how to make the best use of it though
The way I made the most of my L20.5's was by matching output transistors and then setting IQ to approximately 5ma per transistor. Measuring the voltage over the 0.05 ohm resistors. It was not fast work, had to try many possible good matching transistors before I got the current to split evenly over all output transistors. Was surprised when it finally did
The way of Kozard.
The way I made the most of my L20.5's was by matching output transistors and then setting IQ to approximately 5ma per transistor. Measuring the voltage over the 0.05 ohm resistors. It was not fast work, had to try many possible good matching transistors before I got the current to split evenly over all output transistors. Was surprised when it finally did
The way of Kozard.
Last edited:
Hi Berlusconi
I have a few hours ago finished repairing another pair of the 10 x L20.5 boards that became damaged in use over a year ago. This time instead of replacing everything I have just replaced what was damaged - just for info all of the power transistors, the 68ohm, R050 resistors and the 1k resistor with the 953ohm resistor. I am listening to them now and they sound fine in fact as good as the previous boards where I replaced all of the components. However, I have a concern with one of the boards the millivolts only measures 10mv and runs very cool. The other board measures 20mv and runs a little warmer, I hasten to add I have changed nothing else and I suppose what my question is will it be okay running at 10mv. Bill
I have a few hours ago finished repairing another pair of the 10 x L20.5 boards that became damaged in use over a year ago. This time instead of replacing everything I have just replaced what was damaged - just for info all of the power transistors, the 68ohm, R050 resistors and the 1k resistor with the 953ohm resistor. I am listening to them now and they sound fine in fact as good as the previous boards where I replaced all of the components. However, I have a concern with one of the boards the millivolts only measures 10mv and runs very cool. The other board measures 20mv and runs a little warmer, I hasten to add I have changed nothing else and I suppose what my question is will it be okay running at 10mv. Bill
I think very cool is a bit too cool. Check if the current is shared by measuring if the voltage over the surface mounted resistors is the same between them. On one of my very cool running boards there was no voltage at all over two or three of them so I increased Iq till it was. At that point I had matched transistors on there but I don't know if that's critical actually.
Hi Postpunk,
Many thanks for your reply, I have just checked a couple of the R050 resistors and it would appear you are correct one of them measures 0.6 mv and the one next to it measures 0.0 mv's. I am sorry but I am not particularly knowledgeable with regards to terms (IQ) or how to increase the voltage across the resistors.
Many thanks for your reply, I have just checked a couple of the R050 resistors and it would appear you are correct one of them measures 0.6 mv and the one next to it measures 0.0 mv's. I am sorry but I am not particularly knowledgeable with regards to terms (IQ) or how to increase the voltage across the resistors.
Measurements with a multimeter in this region have very little accuracy and the measured value isn't of importance because you can not adjust it. You should realise that potential drop over resistor is consequence of current and that current can be adjusted elsewhere in the circuit and only with the missing potentiometer. You look like a pilot in a plane without a throttle. He can just observe the decrease of altitude. Simply, LJM has omitted potentiometer for the reasons known to him. We can just speculate why he did that. Just install a potentiometer and that's it. As simple as that. And don't take all LJMs words at face value. He is a "wise" guy if we neglect some important issues that matter, at least to me.
Last edited:
I do agree with you Berlusconi, and I find your posts useful and inspiring at all times. One has to read what LJM says and what he does not say very carefully. He said there is almost no difference if QC is 100ma or almost no QC at all. Can it be that 100ma would be the theoretical best value? Shared by the 8 transistors that is. That would be in line with your measurements, no?
Hello postpunk
Native Chenglish speakers have a huge advantage over the rest of us. They may read scientific papers easily and use them if they need and then, suddenly, when faced with inconvenient questions they may decide that they don't understand a single word. A good Chenglish speaker may even claim that Douglas Selfs' designs are his own but he has lost all his schematics. Where? How? Have they burnt with all classical texts during the Alexandria Library Arson, in 48BC.
However, we may manage to resolve all our dilemmas. A hint: just look at the only accurate fragment of the schematics, the output stage, and apply:
1. Ohms' Law
2. The First Kirchhofs' Law
It's easy. And forget "per pair" nonsense.
But we want the answer to another question, don't we?
Last edited:
We have spectrum analysers at work and one of them works all the way down to the audio band. I am not sure how to make the best use of it though
...
That is exactly the way to go. Simply measure the THD value as a function of either quiscent current or voltage drop over R13 or R20. In fact, it doesn't matter at all which variable you use because they have essentially the same character - they represent current through your output transistors in the quiescent state. Your optimization variable is simply position of the (missing) potentiometer. Hint: you can't measure quiescent current and THD simultaneously because the signal used for THD measurements would affect the measured value of the quiescent current. The procedure is as follows: (1) turn the base signal of and measure the quiescent current (2) turn the base signal on and measure the THD+N. Repeat the process for different potentiometer settings. Important: use a clean signal because bad signal source may introduce own distortion.
Over the past few weeks I have also built a THD measuring circuit and I use Arta program for measurement with my built in Realtek sound card. That approach works and is much more usable than qualitative, subjectivistic approach. Currently I use it to measure my Accuphase 60 clone but as soon as I finish that, LJM boards come to the bench - to face reality.
That's exactly the way to go. Building an amplifier from kit is like Russian roulette: you never know when disaster would strike. I prefer to measure every component before and after soldering and purchasing components from reliable source. I purchase from Farnell, Mouser and Reichelt. Reichelt is a great source for generic semiconductors.... The way I made the most of my L20.5's was by matching output transistors. ... Was surprised when it finally did
The way of Kozard.
Finally, let me explain what should be measured to establish the quiescent current.
First look at the schematics below, it is published in the post #3 of the present thread, by LJM himself.
Now observe that all currents from Q5, Q2, Q14, Q18 and Q20 flow into the resistor R13. This simply means that the quiescent current goes through the R13. When you measure potential drop over R13, you have just to apply Ohms law to establish the quiescent current. However, for practical reasons, voltage drop is more convenient value .
Yeah but if the transistors in parallel don't turn on at the same time won't that make the transistor that
is on hotter causing them to amplify even more and ultimately do all the job? That is why I measure the voltage drop over the emitter resistors. To make sure that they are all turned on at the same Ruhestrom. I do have them Potis installed.
On my matching components boards the resistance required for that is 1050 ohm on both boards. I am sure that Kirchhof was right that current adds up. Still, if I have four transistors per rail I want them to share duty. Otherwise I would use just one for simple action.
Thanks for the hints on Specki usage. I have only a very clean 10 mhz reference signal available. Peak. I don't think that will be usable for THD measurements of audio amps though. What would you use?
is on hotter causing them to amplify even more and ultimately do all the job? That is why I measure the voltage drop over the emitter resistors. To make sure that they are all turned on at the same Ruhestrom. I do have them Potis installed.
On my matching components boards the resistance required for that is 1050 ohm on both boards. I am sure that Kirchhof was right that current adds up. Still, if I have four transistors per rail I want them to share duty. Otherwise I would use just one for simple action.
Thanks for the hints on Specki usage. I have only a very clean 10 mhz reference signal available. Peak. I don't think that will be usable for THD measurements of audio amps though. What would you use?
@postpunk
Arta program has own waveform generator which can be used through the sound card output.
Regards
PS: Don't worry about possible differences among the output transistors. These 0R1 resistors are there to take care of that. If there are too large differences among the output transistors you will see that as a difference in temperature, but I think this output stage handles the differences well, especially because you have sellected similar transistors.
Arta program has own waveform generator which can be used through the sound card output.
Regards
PS: Don't worry about possible differences among the output transistors. These 0R1 resistors are there to take care of that. If there are too large differences among the output transistors you will see that as a difference in temperature, but I think this output stage handles the differences well, especially because you have sellected similar transistors.
Last edited:
To measure what you read on your scope, are you using a square wave input to the amplifier? At what frequency? What is your resistive load on the output your measuring across? Curious