Exploring the monster (trafoless mic pre input)

Hey there.

This is a means to an end thread. =)

I want to explore a statement that Winston made in the trafoless mic pre section in regards to the Le Monstre Design. I'm not a fully qualified circuit bender yet, but I tend to learn best when I have specific application in mind.

Here's the quote:
[quote author="OOO Boogey"]
Brad, have you played with Jean's 'Le Monstre' circuit at all? I have a couple of small-signal line amp circuits I built that were 'inspired' by this. One is single-ended but has the same elements as Monstre - BJT cascoding J-Fet's (// array in this case) that is fed into the so called 'Darlingnot' - and it is sonically the most tube-like squalid-state circuit I've cobbled together thus far. Harmonic Analysis is almost spot on that of a good triode valve.
.........
I have to say that I'm preferring the single-ended version that I built and this does not have the same issues with matching. Again, this amp consists of a parallel set of 2SK170BL's (4 in this case for 6dB theoretical noise improvement) with a single low-noise transistor as a cascode - a Fairchild KSC1845[/quote]

The link I found for The Monster design was at:
http://www.tcaas.btinternet.co.uk/hiraga.htm

My specific application is an 8X1 mixer. I've got an API 325 card that I plan to use for the combining network, which sets my power supply voltages at +/- 18V.

So the little static electricity discharge in my head says that the above could be a decent candidate for the 8 balanced mic inputs for this project. That is using the input stage, and cutting out the darlinnot stuff.

My main problem is that in electronics, I can't pass the blank piece of paper test yet ( this is an interview technique I use for software developers, which is simply hand someone a blank piece of paper and ask them to write a program =) ).

If I could get some help with a drawing of the above statement ( I get in big round concept what he's saying, but if I were to attempt a drawing of it, y'all would laugh your asses off), I'm more than happy to do a board layout for it so that the group can have another circuit to experiment with.

Regards

ju

The trouble is that neither the Hiraga or the Winston circuit described (as far as I can decipher his statement) have balanced inputs that would lead straight to making a mic preamp.
Of course, there are known ways to hook up opamps to make a mic preamp (for example: instrument amp, PRR's balanced input, etc). A quick search on the board will turn all of them up.

Ahh.. a good point made there :grin:

yes, I could easily use a differential opamp design( and can even draw that myself =) ) or one of the many other designs available on the board, but thought Winstons comments might garner some additional exploration.

Regards

ju

the problem facing you is that using a multi-opamp receiver circuit only adds complexity and color to a design. unless you are using the "monster" strictly for sonic charachter(in addition to the opamps you have used to unbalance..)you have now compromized it's sound with additional parts. you might as well just add another opamp and make it a 3 opamp preamp...

You can take two colored-with-second SE preamps and combine for a balanced config without too much trouble, but the irony is it will wipe out your fat etc. coloration from all the second harmonic, to the extent that the two halves are matched!

Since third won't be affected, the result may be less than pleasant.

...

Is it possible to make a balanced input buffer with low noise and good CMRR? Then the gain could be supplied by warm and fuzzy circuits like Hiraga's Monster.

So in reviewing most of the designs(transformerless discrete differential) that have been discussed on board(search engine), most of them are using a single ended supply (PRR's 48V design, TK's design (tho he was working on a +/-24V design circa a year ago). What are the design implications of trying to adapt something to a +/- supply?


I'm building up a 5534 input design, to have something to compare against, I'm still diggin around, if I'm missing some obvious things to explore, I'm very open to suggestions =)

ju

Adapting to +/- is no biggie.

To Tamas's question, sure---a very clean diff front end is what I would tend towards, and then add the coloration later. Despite the addition of parts I sense that would be the most versatile, as you could switch to a different middle section as desired.

Brad,

Can you recommend a type of front end that you would consider using in a multistage circuit?

Ju,
I don't mean to derail your thread. Let me know if you want me to bug out.

Thank You,
Tamas

Oh no, you're fine here. Seeing an answer to your questions may help me with my own =)

ju

The tricky part seems to be keeping the front end fairly low gain WHILE making the feedback network values small for low noise. I have a Projet-66 like input stage in mind using cascoded power transistors and pretty high current. I hope to use power transistors for low noise and cascoding to reduce input capacitance. Need time to work that out using PSPICE.

...

Winston.

Thanks very much, got it.

Here's a contrasted version, if it's easier for people to read:

http://www.shinybox.com/diy/Winston.jpg

Regards

ju

Re the diff stage: what gain would be low enough? I guess we want to tolerate about +/- 1.1V differential, i.e. 0dBu, based on what I have been hearing from peoples' experiences.

We would want to get out of noise enough to make the next stage's noise fairly small by comparison. But the constraint on the overload behavior of the next stage is then that much worse.

If we do a differential-voltage to single-ended-current conversion we may be able to then develop a voltage at the input of the single-ended stage that can be varied with a resistor without greatly affecting the S/N ratio. But this is in essence what is going on with the traditional differential current-feedback configurations using Sziklai pairs or more complex versions, except that many times the outputs from the composite "collectors" go on to op amps.

[quote author="tk@halmi"]The tricky part seems to be keeping the front end fairly low gain WHILE making the feedback network values small for low noise. I have a Projet-66 like input stage in mind using cascoded power transistors and pretty high current. I hope to use power transistors for low noise and cascoding to reduce input capacitance. Need time to work that out using PSPICE.[/quote]

Weren't we in a similar situation when we were trying to boost ribbon microphone output using phantom power? This is a great thread:


http://www.groupdiy.com/index.php?topic=5743&postdays=0&postorder=asc&start=60

[quote author="PRR"]
Transistors should be big for low noise. It will "work" with little transistors like 2N3904 but if it is noisy try 2N4401/4403/etc big fat-Base switches before you condemn it. LM394 or the MAT or THAT similar parts are also good bets.
[/quote]


If I apply similar( but potentially incorrect :grin: ) logic, would employing said fat-Base switches in an arrangement like:

http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/part1/page3.html

Get me closer to disco?

Regards

ju

I've been reading a bit more on differential amps, and see that adding a current source in place of the tail resistor will reduce common mode gain significantly.

In selecting the transistor for this current source, is there anything special to take into consideration? If I'm using 2n4401 for the differential pair, is there a reason not to use the same for the current source?

Also in this vein, I'm wondering what kind of precautions are normally employed to prevent base-emitter reverse bias problems.

I'm assuming this is addressed using inter-emitter resistors?

Again, I'm really in my infancy as far as this goes, and this post is really the result of spending a couple hours digging and reading.

I've got a simple schematic I'll post when I'm back at the studio, but thought I'd rattle off some of my current thinking.

Regards

ju

[quote author="fum"]...
In selecting the transistor for this current source, is there anything special to take into consideration? If I'm using 2n4401 for the differential pair, is there a reason not to use the same for the current source?

Also in this vein, I'm wondering what kind of precautions are normally employed to prevent base-emitter reverse bias problems.

I'm assuming this is addressed using inter-emitter resistors?...

Regards

ju[/quote]

Nothing particularly special---make sure the breakdown voltage (the Vcb rating is close enough if your base bias network is appropriately low impedance) is adequate for the maximum signal swing in common-mode.

The output Z of a bipolar current source limits out at several megohms for large emitter resistors, but then those mean you are sacrificing signal swing. Usually the two-diode bias is adequate for a long-tailed pair I source, and also has a tempco that tends to stabilize the stage gain for directly-tied emitters in the pair.

Standard-red LED bias is good for a stable current OTOH.

Then there are a number of higher performance topologies using more transistors---but usually not necessary.

As far as Veb breakdown, the desire is to never have any substantial reverse current flow at all---it degrades beta and noise performance. Often individual diodes are placed in reverse across each b-e junction. Sometimes diodes are put from base to base of a differential pair---although this restricts slew rate in some cases.

> hand someone a blank piece of paper and ask them to write a program =) ).
> adding a current source in place of the tail resistor will reduce common mode gain significantly.

Don't condemn the simple resistor tail until you try it.

At 15V supply, you can find 53dB CMRR in a simple resistor, and you generally can't actually get that good in real life with real cables and real microphones. 20dB CMRR is a heck of a lot better than none, and I did a lot of work with 0dB CMRR.

I quote the Beatles:

"Gotta good reason, for takin the easy way out, now" =)

thanks

ju