[NelsonVandal]

I finally boxed my JISBOS amp. It's gone through some changes. At first it was 3 ch active ground, but because of massive current draw and some stability problems, it's now 2 ch with TLE2426 as virtual ground. The sound was better with active ground, better mid frequenzies and better center information like lead vocals. This is of course unity gain, no voltage amplification. Absolutely noise free. It's not an exact JISBOS. I use BC549/559 because I had better matched pairs and MJE243/253 instead of BD137/138 because I like them better.

It's powered by ten 950 mAh AAA's or an external stripboard 24 V TREAD. 12 V is the lowest voltage supply without any obvious degradation in sound. It's designed for 30 V. There's a built in trickle charger. At 24 V it get's quite hot, a perfect hand warmer. I haven't measured the current draw but it should be something like 100 mA. DC offset is trimmed to about 1 mV.

It sounds very clear and transparent, fast, smooth, warm etc and puts all my opamp based amps to shame. I haven't yet done any direct comparisons to open loop "diamond buffers", but I'll later report how it sounds comared to LISAIII buffers. I can't imagine anything sounding much better than this.

I prefer board mount jacks. I wish I used it on my other amps.

A bit of a tight fit when connected to TREAD.

Hammond C1201, would have been prettier black, but I had no patience to wait for it.

Used to be 3 ch, that's why the strange layout and mutilated LM317. No coolers, hope it wont start a fire

.

Eight batteries this way, two in the back the other way. Very tight fit.

 

[amb]

Using a single supply and without a ground output channel, your TLE2426 becomes a return current bottleneck. For 2-channel operation it would be much better to ditch the TLE2426, rewire the batteries to two groups of 3, and use the middle as the ground (to make a split supply). Unfortunately, this means you need a DPST switch for power...

 

[NelsonVandal]

Thank you for your suggestion. I really hope this can improve the sound (even though there's not much to improve on). I have an OPA551 lying around. The amp will never draw more than 200 mA. Do you think replacing the TLE with this would be as good as using a split supply?

 

[amb]

If you have room for another opamp, then OPA551 would definitely be much better than the TLE2426 alone. You could use the TLE2426 to create IG and then buffer it with the OPA551 in unity-gain configuration for OG (Mini³ v2 style, only with different chip) to make a 3-channel amp.

 

[cfcubed]

Wow, impressive build - You must really have wanted a JISBOS amp on the go

Quote:

Originally Posted by amb /img/forum/go_quote.gif Unfortunately, this means you need a DPST switch for power...

Anyone see my other posts must think I have relays on the brain, but I think a low-voltage DPST/DT relay for power-on (switched by your ALPS & 1 batt side) could preserve the nice on/off pot.

 

[majkel]

Nice work!

I'll add my two cents. Does it make sense to build 24V driven unity gain amp? The highest voltages to obtain from a source are around 2V RMS (CD/SACD line out). This means the maximum amplitude is 1.414V, and the rest of 12V per signal half is wasted as the heat dissipated by the transistors. I assume there is no voltage amplification stage just the discrete buffers, and that's why. All the thing is that you still cannot drive high impedance headphones better than the source.

 

[TzeYang]

majkel,

a redesign in terms of parts selection comes to mind.

Would either you or amb do it? (prays hard for something that is unlikely to happen)

 

[NelsonVandal]

Quote:

Originally Posted by amb /img/forum/go_quote.gif If you have room for another opamp, then OPA551 would definitely be much better than the TLE2426 alone. You could use the TLE2426 to create IG and then buffer it with the OPA551 in unity-gain configuration for OG (Mini³ v2 style, only with different chip) to make a 3-channel amp.

I've added the OPA551 as rail splitter. Placebo is a powerful thing especially for us DIY'ers, but I think the mid and center information is better and there's an overall warmer tone, more like when I had active ground. I'm very happy with this amp. Thank you again Amb. If only I could have this sound in a portable with gain, then there would be no need to look for anything else.

I don't want to use OPA551 as active ground, because I don't think it's good enough. I've tried this many times with open loop buffers - opamp or opamp/buffer in closed loop as active ground. None of the opamps I've tried is good enough. The best are AD829 and AD744 but like I said not good enough, and that's why I wanted to try JISBOS. The opamp used in active ground has such an impact on the sound, just as much as left and right.
Quote:

Originally Posted by majkel /img/forum/go_quote.gif Does it make sense to build 24V driven unity gain amp? The highest voltages to obtain from a source are around 2V RMS (CD/SACD line out). This means the maximum amplitude is 1.414V, and the rest of 12V per signal half is wasted as the heat dissipated by the transistors. I assume there is no voltage amplification stage just the discrete buffers, and that's why. All the thing is that you still cannot drive high impedance headphones better than the source.

Like I said, this amp doesn't sound good with less than 12 V supply. If there was an amp with this SQ powered by 6 V or something like that, I would be thrilled. All chip-amps sound faulty compared to this one.

My CDP can't drive phones from line out, not even high impedance ones. Believe me I've tried. It's an old all discrete Harman/Kardon. Maybe the new opamp based CDP's can drive phones from L/O, I don't know. I often use line out from iRiver IHP120 to HD650, and it sounds very good - better than with any of my other amps.

This is how it looks with OPA551 as rail splitter.

 

[NelsonVandal]

Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif majkel, a redesign in terms of parts selection comes to mind. Would either you or amb do it? (prays hard for something that is unlikely to happen)

I've already asked Amb this on Headwize.com. It's not on his priority list. An amp like that would beat all competition as a portable amp for IEM's or low impedance phones. Like I said, chipamps sound like bubble gum compared to buffers only amps.

TzeYang, think of something like this amp balanced, 6 dB gain. It would be the amp for us HD650 fans. I wish I had a balanced source.

 

[NelsonVandal]

Please remember this, I have a stripboard LISAIII clone, not the original. There's some differences between mine and the original, like different output transistors and a bit lower capacitor values. The original is probably better. Still this is the best chipamp I've heard.

I compared the JISBOS amp to the LISA clone with bypassed opamps in L/R and AD829 comp pin out as ground, and then with the whole ground channel bypassed. Kind of a shootout between JISBOS and "the best" diamond buffers. It uses the same transistors BC549/559 and MJE243/253. Listening is done with HD650.

The comparison isn't totally fair. In the LISAIII there's a TLE2426 as rail splitter, but the current draw for 2 ch is about 18 mA, so the TLE should be sufficient. To LISAs favour the supply is 18 V. There's a substantially improved sound in LISA with L/R opamps bypassed.

The JISBOS amp is the winner. Better details, clearer soundstage with everything in the right place. The diamond buffers are very close, but the overall sound is slightly "chaotic" in comparison, and the center information is a bit vague. The latter is improved when using active ground. An output impedance issue?

 

[kvant]

Nelson,

I am really interested in your experiments with buffer-type amps as it seems those are the way to go for me as well. A have a few thoughts though:

I think you underestimate the influence of virtual ground in single supply designs. You carefully treat the current going from one supply rail to one side of the headphone driver (you have a nice buffer there), and then you sink the same current from the other side of the driver through OPA (or TLE in the worse case) to the other rail. This looks like a bottle-neck to me and a clear departure from your opamp elimination strategies. Also, I do not see why OPA as a rail splitter should be better than the same OPA as a part of an active ground channel.

As for the comparison with LISAIII buffers - don't they switch to class B with their only 18mA of (I assume quiescent) current draw? That could be their another disadvantage when they are removed from the opamp's feedback loop that would correct for the increased distortion otherwise?

BTW, one more thing to my first point. The fact that the LISA buffers draw only those 18mA "at rest" does not mean that the current through the load is not larger and that the TLE does not get overloaded. Actually, the quiescent current does not even reach the virtual ground pin of the TLE.

 

[majkel]

Oops, I didn't expect the source is desktop CD player, not a portable mp3 player. Sure, line out is not dedicated for driving headphones but I thought you used a portable amp for portable sources having headphone outs as well.
Regarding ground channel - this is true what kvant says. I always use a rule that the ground channel current efficiency has to be twice the channel efficiency because you sink both channels to the ground channel. It's logical IMHO.
Redesign... I'm currently crazy about my designs. Wanna exhibit something on a local show in early November. However, I succeeded once on another forum by telling a guy what circuit to use which I never tried and got very good feedback. It's all the way high current op-amp based and might be built with 2 pieces of AD8397 but the guy told me 2x LT1210 acts better on channels, AD8397 remains on the ground buffer.

 

[fortney]

N.V.--Have you built a Szekeres orPoormans Zen by Sijosae (Also current buffers using MOSFETs) and if so can you compare to your new creation? TNX

F

 

[NelsonVandal]

Quote:

Originally Posted by kvant /img/forum/go_quote.gif Nelson, I am really interested in your experiments with buffer-type amps as it seems those are the way to go for me as well. A have a few thoughts though: I think you underestimate the influence of virtual ground in single supply designs. You carefully treat the current going from one supply rail to one side of the headphone driver (you have a nice buffer there), and then you sink the same current from the other side of the driver through OPA (or TLE in the worse case) to the other rail. This looks like a bottle-neck to me and a clear departure from your opamp elimination strategies. Also, I do not see why OPA as a rail splitter should be better than the same OPA as a part of an active ground channel. As for the comparison with LISAIII buffers - don't they switch to class B with their only 18mA of (I assume quiescent) current draw? That could be their another disadvantage when they are removed from the opamp's feedback loop that would correct for the increased distortion otherwise? BTW, one more thing to my first point. The fact that the LISA buffers draw only those 18mA "at rest" does not mean that the current through the load is not larger and that the TLE does not get overloaded. Actually, the quiescent current does not even reach the virtual ground pin of the TLE.

As you understand, I don't know much about electronics. That's why I'm glad for your replies and the posts of more experienced DIY'ers. I really don't understand all this about grounding - sinking, sourcing etc. I think using an opamp between IG and OG (active ground) like in PIMETA, PPA, LISAIII etc has a major influence on the sound. I've never thought of this when using an opamp as VG/railsplitter. Are you saying this affects the sound in the same way as using the opamp as active ground?

How is this problem solved in commercial two channel amps like RSA, Meier etc?

Do you think it would be better to use 78xx/79xx or splitting the batteries. Is it safe to split the batteries? What happens if one half drops faster than the other? Wouldn't there be a massive DC-offset? Should I try to get it to work with a JISBOS as active ground again?

I think I have a good ear for Hi-Fi, and I'm really pleased with this "amp". It's way ahead of my opamp based amps. I'd be thrilled if there's room for improvement.

 

[NelsonVandal]

Quote:

Originally Posted by fortney /img/forum/go_quote.gif N.V.--Have you built a Szekeres orPoormans Zen by Sijosae (Also current buffers using MOSFETs) and if so can you compare to your new creation? TNX F

No, I'm really want my amps portable. I almost always listen moving around in the house or doing the dishes or cleaning up or something, but I think I'd like a simple MOSFET amp like the Szekeres. I'll probably try it sometime.