Another Discrete Amp - GainBloak

[Svart]

Oh i can't wait to give them a try!

 

[bovox]

Nice amps, I´m certainly gonna give them a try, but...
A couple of question´s; Was the J-fet type stable down to unity gain?
Your scheme had a gain at about 26 dB.
And why did you change Q2 to 5087 instead of 560?
Have you measured the input imp. with the BJT version with 550 ?
Or is the calc. R10 at 3 k just to imply that its best with lo source imp. = less noise. Thanks !
Cheers Bo

 

[tk@halmi]

[quote author="bovox"]Was the J-fet type stable down to unity gain?[/quote]
It is only the iput transistor that is a FET. All Bloaks can be made unity gain stable by connecting a 10k to 15k resistor between the output and the non-inverting input.

[quote author="bovox"]Your scheme had a gain at about 26 dB. [/quote]
The open loop gain is at least 75dB, but often more. The 26dB to 30dB is a guideline for clean gain. For the picky ones you know... [/quote]

[quote author="bovox"]And why did you change Q2 to 5087 instead of 560?[/quote]
Please ingore that. The BC560C has better fT and it is widely available.

[quote author="bovox"]Have you measured the input imp. with the BJT version with 550 ? Or is the calc. R10 at 3 k just to imply that its best with lo source imp. = less noise.[/quote]
The input impedance is high although I don't have a number for you.
In real circuits, if the output impedance of the previous stage is 5k or less the BC550C input works without creating huge gain dependent DC offset on the output.
Over 5k you want to use the FET input version and that helps minimize the DC on the output.

Here is a link to the latest documentation and the evolution of the Bloaks:
http://www.purpose.freeserve.co.uk/BloakDocs.zip
(Many Thanks to Peter for hosting this. My Verizon website is failing.)

It includes drawings for:
- GainBloak
- FetBloak (2SK170BL input)
- Improved FetBloak (more stable VAS)
- The BigBloak (Darlington output transistors)

The latest one uses the MJE800/700 darlingtons. This reduces distortion by 75% through improved open loop gain, and drives impedances down to 75 ohms. It retais the 100V/uS+ slew rate. This one really consists of eight transistors, but it uses the same PCB as all the others.

Cheers,
Tamas

 

[Samuel Groner]

For the picky ones you know...

I feel adressed! :green:

Thanks for the doc!

What is the function of R2? IMO we could simply skip that one--saves one part, improves CMRR and makes gain setting more predictable (and down to "true" unity gain).

Samuel

 

[bcarso]

tk@halmi wrote: "It is only the iput transistor that is a FET. All Bloaks can be made unity gain stable by connecting a 10k to 15k resistor between the output and the non-inverting input."

??

 

[Samuel Groner]

Do I read your two question marks right that you agree with me that with a 15k feedback resistor we get about 6 dB and not 0 dB? (Horrible, my English syntax...)

Samuel

 

[tk@halmi]

The gain setting is not accurate due to R2. I have found that the amp works with large variations in feedback network implementations when R2 is utilized, partly because it negates the network at low gain settings. This makes it work in most DIY abuse situations independent of load, etc. The value of R2 could be larger, and up to infinity, or eliminated. Brad is right in that it will not be really unity gain. If you need a true unity gain buffer or accurate gain settings you want to use something else. There are plenty of buffers published out there. Numerically the circuit's parameters are modest, and for audio it does what it does pretty well, but it is NOT the solution to world hunger.
Truly, I think everyone should feel free to experiment with it and come up with their own pet circuit implementation. Roll your own, have fun!!!

Cheers,
Tamas

PS.: If you want you can put a tube in it and I promise not to get pissed.

 

[bcarso]

I was drawing attention to the "non-inverting"---I should think you meant inverting.

 

[Svart]

not sure it matters here but what HFE range should the BCXXX parts be in? C seems to be the most common to get but B is MUCH cheaper..

A: 110-220
B: 200-450
C: 420-800

 

[tk@halmi]

I have not built them with anything besides the C parts so I don't have any experience with the effects of using lower hfe pats. The BigBloak has so much open loop gain that it may not matter for that circuit at all. Mouser has a good selection of transistors. They have the BC550CTA and BC560CTA for 6 cents in single quantities, or the way I get them is, a hundred at a time for $4.

Cheers,
Tamas

 

[Svart]

I bought a few of each to do some testing. I don't think the A will work well but B and C should be fine.. we'll see!

 

[tk@halmi]

Here is another little preamp I just completed for a jamming buddy who wanted a compact unit for his desktop studio.
front
open

It is based on the JLM boards and uses two BigBloaks for gain and the signal chain is LL1538XL+BigBloak+BigBloak+CMOB-2H.
When you use the JLM board do not hook up the gain potentiometer as shown on his drawings. His is wired so that as you increase gain the value of feedback resistor increases. This is bad for a current feedback amp as you limit the bandwidth as you increase gain, just the opposite what you want. I recommend that you wire the pot in a more traditional fashion as shown on the document packet posted earlier. I used a 10k dual rlog also from JLM.
I used the 2SK170BL for the input and a 22k resistor for R2 in both units, but it could be further increased for more accurate low gain settings.


Measurements for the preamp, with 600 ohm termination resistor on the output, using RMAA were the following:
- Loopback test THD of the soundcard is 0.001% with only second and third harmonics above the noise floor.
- At 50dB gain, 0dB output level the THD is 0.008% with only second and third harmonics above the noise floor.
- At 67dB gain (max gain), +16dB output level the THD is 0.025 with mainly second and third, but a little fourth and fifth harmonics showing above the noise floor.
- Bandwidth is 5Hz to 48KHz (upper measurement limit of my soundcard in 96KHz mode).
- Minimum gain is 28dB and the max is 67dB.

He promised to make some samples of his accoustic guitar, snares and hi-hats during the Holidays. I will be mostly out of computer range starting this Thursday, for a week.

Cheers,
Tamas

 

[clintrubber]

[quote author="tk@halmi"]
The latest one uses the MJE800/700 darlingtons. This reduces distortion by 75% through improved open loop gain, and drives impedances down to 75 ohms. It retais the 100V/uS+ slew rate. This one really consists of eight transistors, but it uses the same PCB as all the others.

Cheers,
Tamas[/quote]

Ha, 75 ohms ! :wink:
There was this thread recently that quickly tuned into overengineered G-Pultec-EQ-drivers to address the input-impedance down to 75 for some mid-to-higher freq.
I'll be using PRRs Tori-amp there, indeed, overengineered, but why not - but I see this Bloak could be used there as well. Could fire one up, see how it compares.

Bye,

Peter

 

[clintrubber]

[quote author="tk@halmi"]
- The BigBloak (Darlington output transistors)

The latest one uses the MJE800/700 darlingtons. This reduces distortion by 75% through improved open loop gain, and drives impedances down to 75 ohms. It retais the 100V/uS+ slew rate. This one really consists of eight transistors, but it uses the same PCB as all the others.

Cheers,
Tamas[/quote]

Hi Tamas,

Some more on those Darlingtons: couldn't find those types, but found BD678A & MJE802 (just a different voltage rating, so fine), which I should both be able to get.
It'll be not very critical I expect, but perhaps still any other suggestions
for the MJE700 ?

For the 1N5298, I found the J505, at 1.0mA, or could wire up a JFET myself.

Regards,

Peter

 

[tk@halmi]

Peter,

I think any of the BD677/678, BD679/680, BD681/682 pairs would work if you can find them.
I used a KSK30 (2SK30 equivalent) that works for the current regulator just connecting the gate to source. There are many JFETs that have Idss ratings around a milliamp.

Tamas

 

[clintrubber]

Hi Tamas,

[quote author="tk@halmi"]Here is another little preamp I just completed for a jamming buddy who wanted a compact unit for his desktop studio.[/quote]
Nice unit !

It is based on the JLM boards and uses two BigBloaks for gain and the signal chain is LL1538XL+BigBloak+BigBloak+CMOB-2H.
When you use the JLM board do not hook up the gain potentiometer as shown on his drawings. His is wired so that as you increase gain the value of feedback resistor increases. This is bad for a current feedback amp as you limit the bandwidth as you increase gain, just the opposite what you want. I recommend that you wire the pot in a more traditional fashion as shown on the document packet posted earlier. I used a 10k dual rlog also from JLM.

Looking at circuits like this, I see most using the gainpot just as you describe them, but I may be getting things wrong.

You're using the dual gain-pot (one section for each opamp) as well ?

I think any of the BD677/678, BD679/680, BD681/682 pairs would work if you can find them.

Thanks for the info, that'll broaden the options. I understand that for a dual section-pre I'd actually only need one BigBloak, or did you use two anyway for that lower THD ?

I used a KSK30 (2SK30 equivalent) that works for the current regulator just connecting the gate to source. There are many JFETs that have Idss ratings around a milliamp.

Right, why go out & pay an Euro for a 10% or 20% JFET-part (sloppy for some reference-ish thing) when you can stay inside & test a few JFETs yourself :wink:

Thanks,

Peter

 

[tk@halmi]

[quote author="clintrubber"]Looking at circuits like this , I see most using the gainpot just...keep his high frequency hearing in top shape.

 

[clintrubber]

[quote author="tk@halmi"]I must have remembered wrong. I did not recall having that short betwen pot terminals 2 and 3. Most likely, I am going nuts. Or just need a vacation... [/quote]
All fine & clear now - and it saved you some PCB-mods.
Yes vacation ! It's near now, just a few more days. Who knows some time to finish some DIY-stuff :thumb:

[quote author="clintrubber"]You're using the dual gain-pot (one section for each opamp) as well ?

Yes. [/quote]
Sorry for asking - I checked your pics a bit closer and saw the two green connection-blocks.

[quote author="clintrubber"]I understand that for a dual section-pre I'd actually only need one BigBloak, or did you use two anyway for that lower THD ?

Both of them were BigBloaks. The new owner likes his sound squeeeeeeeeeeeky clean. This is a guy who gets his ear canals flushed every few months to keep his high frequency hearing in top shape.[/quote]
That sounds serious ! They'll be clean those TXs I guess, but is he totally OK with the thought of those chunks of iron in the signal path ? :wink:

 

[clintrubber]

Wondering a bit about the remark w.r.t. DC-output-offset & what to use as the first device (JFET or BJT).

Above 5k (say 200 Ohms*ratio*ratio) the JFET is suggested. But as I understand it: if a BJT was used then TX-secondary would still provide a low-Ohmic path for DC to keep the DC-output-offset low.
In addition, there's also the +1V offset you mentioned, so the basecurrent of the input-NPN times the Zsource-DC wouldn't change that much I expect.
I might be missing something here of course, so please correct me then.

Or are you using AC-coupling between TX-sec & the amplfier ?

Regards,

Peter

 

[clintrubber]

[quote author="tk@halmi"]Here is another little preamp I just completed for a jamming buddy who wanted a compact unit for his desktop studio.

It is based on the JLM boards and uses two BigBloaks for gain and the signal chain is LL1538XL+BigBloak+BigBloak+CMOB-2H.
[/quote]

Some more questions, hope you don't mind: what's the quiescent current of the BigBloak ? More specific: how much for the power-devices ?

Given that desire of your buddy for a super clean amp that story about a certain optimal amount of mVolts over the 10 Ohm emitter resistors came to mind; IIRIC it applies since it's not a fully class A-amp.
I was just curious whether that thing had been addressed too. With the figures you already obtain it might be more academical than that it could really polish those THD-figures further I guess.

Bye,

Peter