BL range should work best
but read this: https://www.diyaudio.com/community/threads/beyond-the-j-fringe.370957/
but read this: https://www.diyaudio.com/community/threads/beyond-the-j-fringe.370957/
Good read, thanks. Makes a few things clear. Linear Systems both spec Idss for Vgs 10V. So why does testing run to 12V? I am thinking DIY Audio store's article on testing was on advice from NP, as they also indicate 12V. How can Idss values be interpreted correctly when Vgs is outside the specified manufacture's value?
I would also like some feedback on the other question regarding L,R channel differences in matched pairs.
Thanks...again.
I would also like some feedback on the other question regarding L,R channel differences in matched pairs.
Thanks...again.
My interpretation of that pdf is... the parts choice will depend somewhat on the amplifier input topology as well; if single-ended input (buffer), then use the first plot. It doesn't really matter if you're going to use a 4mA or 14mA device... you'll need to know what the (input) impedance of the output stage is.... but the benefits of 4mA vs. 14mA in general, are inconclusive... you can use either device.
With a differential input pair amplifier (Aleph J; the first transistor in a different input pair works as a follower, with a CCS with degeneration), look at the second plot. The plot tells us that the higher Idss parts seem to perform better; however, Aleph J CCS sets the current at 4mA - hence, again, either part is okay to use.... sort of.
Why sort of? Because to me, at higher volumes, the higher-graded Idss JFET will drive the MOSFETs with less distortion... The first JFET (idling at 4mA) needs to produce a fair amount of voltage drive to the output MOSFETs, to generate around 20-25W at 2A bias (JFET current peaks will be much higher than 4mA), so I'd go with 11mA Idss JFETs.
I'd love to see what others think...
With a differential input pair amplifier (Aleph J; the first transistor in a different input pair works as a follower, with a CCS with degeneration), look at the second plot. The plot tells us that the higher Idss parts seem to perform better; however, Aleph J CCS sets the current at 4mA - hence, again, either part is okay to use.... sort of.
Why sort of? Because to me, at higher volumes, the higher-graded Idss JFET will drive the MOSFETs with less distortion... The first JFET (idling at 4mA) needs to produce a fair amount of voltage drive to the output MOSFETs, to generate around 20-25W at 2A bias (JFET current peaks will be much higher than 4mA), so I'd go with 11mA Idss JFETs.
I'd love to see what others think...
Okay, I see where this is going. The bias point for the Class A mirror sets the DC idle, the drive capability is determined by the Idss. So then, how could the spec of the AJ be for 25W as I have seen, I believe this is in the original document? If what you are saying is correct, then these devices should have an Idss of at least 10mA or more. Yes?
Hmmm.... what is "DC idle"?
I will only say this:
The 4mA idle Idss is set to ensure the little JFET can produce the required voltage swing to the output MOSFETs, and achieve 25W power, whilst not exceding their (JFETs) Idss max. Once the voltage swing is there (at low distortions... one would hope), the MOSFETs will provide the required current drive to the speaker coil.
I will only say this:
The 4mA idle Idss is set to ensure the little JFET can produce the required voltage swing to the output MOSFETs, and achieve 25W power, whilst not exceding their (JFETs) Idss max. Once the voltage swing is there (at low distortions... one would hope), the MOSFETs will provide the required current drive to the speaker coil.
Last edited:
When you bias it, you do so with no AC signal. So you are setting the DC point of the input gain stage. That's where it "idles" with no signal, meaning there's 4mA flowing Idss.
In any case, NP's writings are showing an Idss of 4 to 5mA as bias. So I am curious now whether or not that will achieve the full 20 to 25 watts of possible power. Can anyone also chime in on the other question about one channel having a higher Idss pair than the other? (as mentioned earlier in the thread) has anyone done this?
In any case, NP's writings are showing an Idss of 4 to 5mA as bias. So I am curious now whether or not that will achieve the full 20 to 25 watts of possible power. Can anyone also chime in on the other question about one channel having a higher Idss pair than the other? (as mentioned earlier in the thread) has anyone done this?
You bias input differential pair with a DC current, as I explained above. You need to consider the voltage swing that needs to be produced, into a given impedance of the output MOSFET stage, to figure out what the maximum current through input JFET will be. Nevertheless, Nelson has mentioned that even low IDSS devices handle these peaks quite well,... there are no major penalties
The full 20-25W will be achieved easily... you need to pick:
1. the correct idle Ids (set by ccs)
2. the correct Idss of the input JFET to operate them in their most optimal region
... to achieve the lowest distortions
You can use different Idss pairs... but... it seems that for some reason, you do not want to look at that second plot... Once again, look at it and check what distortion levels are achieved, at what voltage swings into what impedance, with 4mA devices and with 11mA devices.
And finally, ask yourself: do I want the same distortion spectre, at all volume settings, on both of my AMP PCBs?
The full 20-25W will be achieved easily... you need to pick:
1. the correct idle Ids (set by ccs)
2. the correct Idss of the input JFET to operate them in their most optimal region
... to achieve the lowest distortions
You can use different Idss pairs... but... it seems that for some reason, you do not want to look at that second plot... Once again, look at it and check what distortion levels are achieved, at what voltage swings into what impedance, with 4mA devices and with 11mA devices.
And finally, ask yourself: do I want the same distortion spectre, at all volume settings, on both of my AMP PCBs?
Last edited:
Sure, sounds good. It's always better to be in the ballpark before installing 8 PCB screws, 4 power devices, soldering a handful of wires before deciding something isn't right. And thus my questions. So based on that, I'll give the close pairs a try using 5mA on one channel, 8mA on the other. We will find out just how audible it is, if at all. Thanks for the feedback.
take these as two pairs, each for one channel
finish the amp, close the lid
listen, enjoy
read the article
then again
then read explanation how actually works Aleph J
then read the article
if i use those two pairs, no explanation needed. you missed the point. it clears one question, the other is more or less resolved. read the article, got it.
🤡
🤡
nope, you missing how Aleph J works
plenty of threads with tidbits covered
just as starter - input LTP is biased with CCS in its tail; while Idss of JFets used is important generally, it is not having much with amplifier performances
in the end, some linearity up or down, but absolutely nothing with power figures
I did short the inputs and measured on no load. O/p bias was set at 350 mv and off set at 0 mv (though it kept moving up and down for both channels by about 1 mv). Left channel rusn hotter by about 2-3 C, Q8 of left being the hottest at 49 C. Here are some pics if that helps.
That's a fairly small case for Aleph J, so high running temps are expected. What will hurt even more than heat is an inability to set the appropriate bias (around 400mV minimum)... because you won't be able to dissipate the resulting heat. Furthermore, the bottom plate is not perforated, so you can expect quite a high internal temperature.
If you want to short the input, you should short differential input -IN and +IN separately to ground...
I think your build is fine; you just need a larger case and a perforated bottom plate...
If you want to short the input, you should short differential input -IN and +IN separately to ground...
I think your build is fine; you just need a larger case and a perforated bottom plate...
so, you're good
there is no (neither significant nor practical) difference in temperature between channels
Hello There, Thank for your reply. Actually its not in the case yet - its a 1 mm plate that I use for pre-assembly and testing before transferring to case. My case is 430x150x415 mm with perforated like you suggested. And heatsink is 50x150x400 mm (approx. 2x6x16 inches) I am hoping its adequate as the temp is holding at room temp in tropics. Noted about the input short - Thank you.
Follow up question, if I may - the idle current draw in is about 1.49A. I am using a 10 A CT transformer and the voltage drops from about 26V DC to 23V on load of 1.49A. Is this normal? For testing bench, its 16AWG all around.
Last edited:
Are you gonna be using a single transformer for both channels? If yes, then the total power consumption is 3A with 350mV bias. Also, keep in mind that with 350mV bias the distortions might be a touch too much... see the first and second posts. The widely adopted bias setting is 400mV... you may like even higher bias, maybe 440mV...?
I expected to see a slightly thicker secondary wire. Aleph J burns around 150 - 200VA, so if you apply a safety margin that most DIY-ers do (x 2), you'll arrive at a 400VA transformer. This will ensure a stable operation (no transformer overheating) even with the bias set at above 400mV.
With all that being said... there's nothing stopping you from using whatever transformer you have at hand.... see how that goes. And see if the transformer runs too hot... then you can simply get a 400VA transformer at a later stage.
I have a 400VA 2 x 20V secondary transformer that gives me plus and minus 24-25 V DC rails with 440mV bias.
If I were to build another Aleh J, I'd go with a single transformer, but separate windings for each channel, with 2 PS PCBs; one for each channel.
Thanks again. Yes, I intend to use a single transformer for both channels. I have a CT 18-0-18 10A and a Dual 18-0, 18-0, 6.3A - latter is much smaller and I was hoping to use that as I read in many forums about dual rectifier being better than a single on a CT. I see from your suggestion to stay higher. What was perplexing me is whether 3V drop is expected with a 4A load on 10A winding - I infer from your post that voltage drop is inevitable (20V X 1.414 = 28V DC, but you are getting about 25V) .
Last edited: