Electronic replacement for a Input transformer?

[bcarso]

[quote author="3nity"]Awesome PRR. that's wwhat i wanted to hear.....
So a transformer makes impedance conversion from primary to secondary to accomodate microphones to a given preamplifiers.....So a Transistor stage does that but not only that but a complete preamp???
[/quote]

Transistors can be made to provide a good noise match for mics directly. Transformers were essential when all we had was hollow-state (as PRR calls it), due to the very high input impedance and fairly high voltage noise of tubes. You could parallel a whole bunch, but they are already heating rooms enough as it is.

Where transformers continue to excel: providing isolation between something at voltage A and something at voltage B, where the difference can be sometimes hundreds of volts. They are coming back into their own in home theater installations, where various cable/rf/video equipment must be referenced to power safety grounds of varying potential. Whitlock told me Jensen is doing a brisk business, without which I guess they would be practically moribund.

 

[clintrubber]

[quote author="bcarso"]Transformers were essential when all we had was hollow-state (as PRR calls it), due to the very high input impedance and fairly high voltage noise of tubes. You could parallel a whole bunch, but they are already heating rooms enough as it is.[/quote]

FWIW, this just came to mind... could be a viable alternative...

There may be 'issues' using this for other types of mics but it's fresh to see a different approach. OK, 'fresh', this one's from 1964 :wink: and if it was truly a better alternative it'd been used a lot more I figure, but still...

And as it seems I guess it could be made to supply phantom right away (i.s.o. the DC for the carbon mic, but I didn't check numbers).

IIRIC we've had a grounded-grid mic-pre design being mentioned here a few years ago as well, from some Russian or Polish folks ?

Regards,

Peter

 

[bcarso]

But it still has high voltage noise, depending on the tube---although a supertriode like the 6C45 wouldn't be too terrible.

An interesting thing about active devices: in principle their noise performance is the same regardless of configuration, if you are getting gain at all. In practice though, the biasing networks required often spoil the non-standard hookup's performance. And as well, if the gain is low the contribution of the following stage is made more important.

In the case shown, the carbon mic needs the bias current to work. Needless to say you would not want your dynamic microphone inserted in its stead. So the bias current has to come from somewhere, and there will be noise associated with it.

As well, most of the time the input Z of a grounded-grid stage would be uncomfortably low for most microphones.

 

[PRR]

There is a cute cheat which might "replace" a step-up transformer.

Take the yellow-plan, wired for gain of 5 or 10, into your hot line input.

You need something more than the yellow plan but far less than the 6-transistor plan below it. Proper biasing, gain-set, and decent input impedance may be hard to get all at once. However with a high-gain "line" input behind it, we don't need to cram the input stage down into the lower quarter of the power supply voltage, and we don't need that high supply rail.

I'm thinking two good-current hi-Beta transistors. 2mA. Fixed-gain, 100 ohms between emitters. 500 to 1K in lower emitter as the output loading. The voltage between the bases needs to be 0.7V+0.2V+0.7V= 1.6V. The bias string current might be 0.2mA. So the resistor between the bases is OTOO 1.6V/0.2mA= 8K, plenty high. Lower collector sits at 1V or 2V, we need that much again for swing, and a few volts more to keep the lower transistor wide-awake. So lower base can sit at 5V or 6V, 1.6V interbase, add a few volts for upper collector drop, need at least 10V total supply. Temperature stability is "poor": a 25 deg C change will shift current to 0.5 or 1.5 times design current; perhaps acceptable in shirtsleeve work. You could add diodes to the bias string but input impedance gets low. You could ponder a Vbe multiplier with a B-E capacitor to get a high AC impedance, but we are really losing simplicity. You could increase the emitter resistor for better tempco and add the R-C gain control to bring audio gain back.

Performance is not great. The ~130 ohm noise resistance is "very bad" compared to a good preamp or even a bad transformer. THD at high level will be non-negligible. It will be even-order and flat with frequency, unlike iron which is more odd-order rising in bass.

==========================================

Grounded Grid just makes no sense as an audio input stage. Or at least has no advantage. Wait: Leach's MC phono preamp does gain bias simplicity from CB connection. But he has a CE version which is like one more part. And the nature of phonodisc work means that you really don't need to get small NF in the electrical system. It has generally been easy enough to get electrical noise FAR below actual surface noise.

> the input Z of a grounded-grid stage would be uncomfortably low for most microphones.

And the noise resistance of a grounded GRID (hot tube) stage is 2.5-3 times the input impedance. So either amp noise is considerable higher than source noise, or source is severely loaded. Either way, the NF looks poor.

==========================================

Here is an observation.

All simple single devices, with both voltage and current gain, semi-suitable for linear amplification, used without transformers, INVERT.

True or false?

A common-cathode stage has both V and I gain, and inverts. Same for its BJT and FET kin.

A common-grid or common-collector stage does not invert. But one lacks I gain, the other lacks V gain.

I have a feeling this is true. I have a glimmer of "Why?". But I may be full of krap.

 

[bcarso]

It would be interesting to find out when the idea of driving the base of the transistor first occurred to Bardeen and Brattain. They were just tickled to get power gain out of the device by driving the emitter.

 

[clintrubber]

OK, grounded-grid... a nice try, and probably fine enough for the original application (carbon),
but definitely enough arguments mentioned to reject it for 'serious audio use', thanks.

Regards,

Peter

 

[CJ]

the hammies like the grid tied down...
but what do they know?
find a woman, you amateur radio guys
leave the 10 meter to the boy scouts

 

[bcarso]

GG ("you with the stars in your eyes....") is great for untrammeled bandwidth. When so much of the physical tube is parasitic C it makes sense.

 

[Dan Kennedy]

A lot in common between common grid (gate, base) and common anode (plate, drain, collector) and wide bandwidth, the voltage swinging aspects are reduced, and all of the parasitics become far less problematic.

Or in the case of stability, more so.

 

[PRR]

> It would be interesting to find out when the idea of driving the base of the transistor first occurred to Bardeen and Brattain.

Perhaps when alpha dropped below 1. Recall that the early devices had current gain in common-base connection, alpha of 2 or 3? Rather than <0.999 like all modern devices in normal operating conditions.

> OK, grounded-grid... a nice try

As a prime low-noise audio amp, it has no virtue. IMHO. Anywhere it works OK, driving the Base should work just as good, with a bit more power gain and flexibility.

Of course we use it everywhere. It's our favorite way to inject NFB in a "simple CE" input. Topside of a cascode to control capacitance or voltage. Flipside of a diff-pair just to get another polarity of signal. Sometimes it is as-quiet as a common emitter stage.

Sometimes it has uses. Imagine the emitter bias is high-Z. With a low-Z mike, you have voltage gain. When the mike plug gets yanked out, voltage gain goes to zero. You are unlikely to have a busted and buzzy mike-line.

I keep flipping on the carbon mike and tube idea. Obviously it is a Cascode. The carbon mike IS an amplifier, moreover a 4-terminal amplifier (two acoustic connections {one usually sealed to sound}, two electrical connections). The tube isolates it from high voltage (which may be the only available supply) and Mu-levers up the signal voltage. Mike signal current is tube signal current (at least if you skip the silly extra parts). Mike signal current is necessarily (for tolerable distortion) much-much less than DC current; same in the tube. So max output voltage is small. But a thumbnail computation says 10V may be possible for a serious close talker, and that will slap a 6V6 pretty well. And simple.

> A lot in common between common grid (gate, base) and common anode (plate, drain, collector) and wide bandwidth, the voltage swinging aspects are reduced, and all of the parasitics become far less problematic.

True. Sure. But also Power Gain is lower.

If you mis-load or dummy-load a common-cathode(emitter) stage to give the same Power gain as its two siblings, can it be just as stable? I think there is some discussion like that in "Designing with FETs" (Siliconex) but I don't feel like bending that far.

Cowles 1967 has some examples of pairs built with various combinations of CE CB CC and shows pretty much the same gross performance can be had several different ways. If you play with plans like some folks do crosswords, this is clear; also that "small" differences end up favoring one or another topology for a specific situation and perceived needs.

 

[barclaycon]

Returning to the original subject of the thread, does anyone have the schematic for Rupert's TLA ? I had a look at the Green schematic - is that the whole story ?
I've pretty much standardized on an MCI design for active balanced input, and the cross-coupled op-amp design that Amek and others use for active balanced output.
There are some interesting circuits out there - Alice use a strange 3 op-amp arrangement in their stuff that seems to work OK.
As far a possible I like to be able to plug things into balanced OR unbalanced gear without 6 dB level drops or overshoot etc.
With electronic balancing you still don't get the isolation that transformers give, but even transformer designs can give problems. I plugged in my EMT 156 limiter the other day to find that the separation between left and right was ridiculous. Looking at the board - they had decided to put compensation caps etc. on the input of the transformer. When being FED from a transformer output it was fine, but any other design - particularly unbalanced, caused very strange effects.

 

[dale116dot7]

So far I like the THAT 'OutSmarts' drivers, they seem to work well without overheating and getting too upset - they seem to balance voltage and current in both legs so it works properly in either configuration. I haven't done that much playing with them in unbalanced mode, because all of my gear has balanced inputs or transformers - I set it up that way for gear that didn't play nice with unbalanced inputs. In theory, you could make an 'isolater' with a balanced receiver and a balanced OutSmarts transmitter, and provided you are within the CM range of the input and output, it should behave roughly as a transformer does.

That being said, I prefer to use CJ-approved parts.... :razz:

-Dale

 

[Guest]

[quote author="JohnRoberts"]

Nobody has bothered to mimic that particular characteristic of transformers (AFAIK).
[/quote]

AFAIK on some frequencies some other principles like a piezo effect are used.

I for one am not sure what it would be useful for?

Better power transfer means less losses that means less amplification factor is needed that means less noises and distortions from active devices.

 

[JohnRoberts]

[quote author="Wavebourn"][quote author="JohnRoberts"]

Nobody has bothered to mimic that particular characteristic of transformers (AFAIK).
[/quote]

AFAIK on some frequencies some other principles like a piezo effect are used.

I for one am not sure what it would be useful for?

Better power transfer means less losses that means less amplification factor is needed that means less noises and distortions from active devices.[/quote]

Power transfer for audio signals was far more useful for tube designs due to higher impedance.

Modern audio power amps outputs are effectively bridging terminations (amp source impedance << loudspeaker), modern mic preamps are also bridging terminations, designed to maximize voltage transfer. Even mic preamps using input transformers are still operating as bridging terminations (voltage not power transfer) and just use the transformers to tolerate higher noise, higher impedance active devices.

Try not to step in any labeled poop.

JR

 

[Guest]

[quote author="JohnRoberts"]

Modern audio power amps outputs are effectively bridging terminations (amp source impedance << loudspeaker), modern mic preamps are also bridging terminations, designed to maximize voltage transfer. Even mic preamps using input transformers are still operating as bridging terminations (voltage not power transfer) and just use the transformers to tolerate higher noise, higher impedance active devices.
[/quote]

You are right, that's why custom DIY and vintage gear is so demanded and expensive... Modern SMD rugs can't be compared in sound quality despite of good measurements...

Try not to step in any labeled poop.

Thanks, it's exactly what I am trying to do.

 

[JohnRoberts]

The last thing I would spend my hard earned money on is somebody else's DIY effort. I appreciate the value of real engineers doing real engineering, and properly managed manufacturing processes.

Not all modern gear is both well built and well engineered, but a lot of old gear is just old gear. I also find test equipment quite useful, but I have been quite clear about my preference for clean, linear, audio paths.

JR

 

[Guest]

The main question is, how to measure the clearness: using sight, or hearing criteria?
:roll:

However, it is much easier to use visual criterion, especially when a bear stepped on an ear... :green:

 

[JohnRoberts]

Null test.. Either it's different or not..

JR

 

[Guest]

[quote author="JohnRoberts"]Null test.. Either it's different or not..

JR[/quote]

Unfortunately, no. A human hearing apparatus is much more forgiving to well visible low order distortions on big volumes, but don't tolerate sharp small errors that are nearly invisible when measured. "The Devil is in details". Even if non-linear distortions on low levels are very small, dynamic distortions screw down an exponential law of decay so the reproduction sounds unnatural. One of my friends was an orthodox engineer until (like me 30 years ago) until recently when I demonstrated him differences in sound of a grand piano in a nice reverberating hall. Now he is a heretic like me. His collection had mostly jazz though hi likes piano. When I brought one nice piano record he said "It is gated!". But it was not... He used nice modern opamps all the way, from top to bottom...

 

[JohnRoberts]

I'm tempted to say once again we agree to disagree, but this is the same old disagreement.

IMO if you can hear distortion, you can measure it. If you can measure it you can manage (reduce) it.

I personally have difficulty with the concept of tolerating or even engineering in modest amounts of euphonic distortions, while simultaneously arguing the evils of some other distortions present at near unmeasurable levels.

If that path doesn't sound better with none of either distortion, you are making an effects unit, not a clean audio path.

JR