Discret balanced line input stage!

Think someone was lookin' for a circuit like this a time ago?
This one runs on a single 24V rail, can also be used as an output driver or unbalanced buffer, drives 600ohm loads / +19dB
The two sweep curves are with 470uF or 1000uF output cap / 600 ohms loading. Distorsion @ +4 dB <0.05% can't measure lower than that for now! Will soon invest in a better distorsion analyzer like the HP8903B or something.
No pcb layout yet! Works fine on a nailboard! I will report any instability when I'm ready with the pcb layout!

Hope this can be to any help!

http://www.vintagedesign.halmstad.net/diy/schem_mods.htm

Hi Tekay!

I was after something like that months ago over at the old place, if that's what you mean. Thanks! What sort of rejection do you get with this circuit?

Bjorn

Ya, do people call you names, or just point and laugh? :wink:

[quote author="Tekay"]No pcb layout yet! Works fine on a nailboard! I will report any instability when I'm ready with the pcb layout!
[/quote]

any plans to make this PCB small and with the 99 pin layout ??
could be a fun option for Project 2
Have you tried this on a simulator yet >... would be intersted to know what voltages are about the circuit, especially the output ... can't be bothered getting a calculator out and I'd only muck it up anyway.

A couple of components do give it a familiar feel and the output pair smell of DIY.
Is that a tant or two I see at the input ? do you have a favourite mic-pre I wonder ?
Why the three IN4148 diodes ?

Did some measurements and the best I could get the CMR was 56dB,
Don't know how good you can get it with simple circuit like this? Will try to improve it later!
I needed to lower the R between the output and base of the negativ transistor from 10k to 8k6 to get it better!

Hi Kev!
Tant's are not nessesary! I started with a Neve BA226 unitygain stage that I used in a custom project!

I wanted to bias the circuit around 10-15mA and with 3 x 1N4148 I landed at 14mA bias!
To improve the slewrate I raised the current thru the driver transistor a little too without instabillity.

I tried it with a API output transformer and it worked fine up to +19dB!

I was thinking off a standard 99 pin layout!
I also have a +/-24V version on the way!

56dB ain't bad at all...rarely do you need more. Looking forward to your +-24V version! Good stuff :thumb:

Bjorn

Thanks Tekay!

I keep trying different output stages at the moment, and this looks like a good one to try.

Cheers!

Mark

I suspect that if you replace the 10k current source resistor with a real current source (a red LED, BC214, and two resistors), your CMRR should be a lot better, I would think. If you add a current mirror arrangement to the bottom of the differential pair (maybe two transistors) you should be able to get better yet again - but then it starts to look like a JE-990 or other op-amp. This basically looks like an 1176 output amp drawn upside down - except that the feedback is through a differential pair - it needs to be to accept a differential input, otherwise the impedances are different in the two legs and you have very poor CMRR then.

The three diodes spread the voltage apart so he can use 22 ohm emitter resistors with a reasonable bias current yet the 22 ohm resistors will likely protect the output devices against a shorted output or that sort of thing. You could use two diodes and something like 4 ohm resistors instead, if you wanted to, I suppose.

Oh yes, an even simpler way of improving the CMRR involves increasing R3 (the 10k from one input to ground) up to about 12.2k. That should help a lot. I wasn't planning on building one so someone else will have to try that one out.

Thanx for the tips Dale!
I will work more with this little gizmo as soon as I find some time!

A 22p from the first transistor to ground will get rid of some overshots!
Still only -0,5dB @ 250kHz!
Think it will be easier to work with the +/-24V version

Hi Tekay!!

Thanks so much for this circuit - I was looking for something like this for a small mixer idea back at the old place. Was going to use ICs but wanted discrete!

This would make the perfect companion to the Neve-esque ouput stage I was thinking of using for mix amps!

GREAT!

Cheers Tom :sam: :guinness: :sam:

If I get some time I will play around with a PCB layout in 990 footprint.....but I'm a novice with layouts. Any major hotspots to pay attention to regarding component placements?

Funny, I just did up a 990 in a 2520 footprint. At these sizes, the parasitics are not a big problem, except for the sensitivity at the output of the first current source or emitter resistor for the input pair. The two input transistors also need to be thermally very close to each other. My 990 layout sort-of followed the schematic and I tested it today - no ringing or anything like that.

-Dale

Hi Guys,

Here is some hints to do improvments on a discrete op-amp circuit.

As me and Tekay have talk about before, this differential amp /op-amp is the first grade of the basic diff.amp circuit with complementary output.

Tekay:s variant have one of all improvment that can be don on the basic circuit and this is the bootstrap for the common emitter transistor, the two 2,2 k resistor with the capacitor to output, (this help the transistor to swing near the bottom end) in the basic circuit this is only one 4,7 k (4,4 k) that set the current trough the common emitter transistor via the bias diodes.

Then you can do a lot of improvments on this circuit in different grade as:

Exhange the bootstrap circuit on the common emitter transistor to a constant current source, this gives a better result than bootstraping.

Use a transistor and a preset resistor instead for the bias diodes so you can adjust the bias/quiescent current in the output transistor more easy.

Make a thermal contact between the the bias diodes/circuit and the output transistors to improve more stable quiescent current in the output transistors, it can be don with a other pair of output transistors connected as bias diodes and fit together with the output transistors

Use a constant current source on the differential input pair emitter instead for the 10 k resistor, this gives better linearity for the input pair.

Add a current mirror on the differential pair collector`s,
this gives better likeness in the input pair and is good for dc-offset and input common mode.

Add a cascade transistor pair on the input pair collector`s and also in combination with the current mirror, this have a lot of influense on the input pair as noise, higher gain, better high frequency response, (in a simple way you can say that the cascade isolate the input pair from the common emitter transistor to avoid the "miller effect" and othet things)

Add a emitter folower before the common emitter transistor base or do a darlington configuratiom of the common emitter transistor, this gives also some of the cascade improvments.

All this gives a lot better measurement quality, but it can be harder to hear the different of one of these improvments compare with the basic circuit.

--Bo

I did a layout yesterday with the 990 pin out! Will etch and test it next time I'm etching!
I skipped the -V pin and the output cap, no room for the cap!
Will post the layout as soon as I tested it!

:green: :green: :green:

Bo Hanson is correct in his list of suggestions on how to improve this circuit.

But without wanting to complicate it further, I would say:-

Yes, change the R3 10k resistor to 12k. R3 is essentially in parallel with the 56k bias resistor, and the combination of 12k||56k is just about 10k (balancing the bridge better, potentially improving CMRR).

Also, remove the 1k resistor from the collector of the inverting i/p transistor. It might make the input pair 'look' more balanced, but it simply serves no function at all.

Alan

did a layout yesterday with the 990 pin out! Will etch and test it next time I'm etching!
I skipped the -V pin and the output cap, no room for the cap!
Will post the layout as soon as I tested it!

Click to expand...

What happened with this litle nice project?
Im looking for something like this, but +/-24 :roll:
Any sugestion about?

I was looking through the adverts in our local paper and noticed that ladies advertise certain services which they say are DISCRETE!!!! Is there a connection here?
Stephen

connection :idea:

Tekay, somebody is making jokes :wink:
Guys, lets talk about transistors not transvestits :guinness: