Input Z question: raising Z before attenuator

I built a few mic preamps of British design (REDD.47) where I included an integral 200R stepped T attenuator on the input (between input XLR and input Tx). While this properly loads the input Tx which wants to see 200R, now that I'm learning more I'm beginning to understand that showing a 200R impedance to most microphones (particularly dynamics) is going to cause enough loading to have major adverse effects. I'd like to increase the input impedance to something on the order of 2k-6k at least.

I do like having the input attenuator functionality (and the preamps do in fact sound quite good with most microphones, even in their current form), but I'm wondering if there is an elegant way to increase the input impedance that the microphone sees. Is series resistance a bad idea?

Thanks!!

Switched pads in preamp front ends can add the series resistance and 200 ohm shunt, only when attenuation is desired. When the attenuation is not needed the mic's source impedance will properly terminate the tx so the extra 200 ohm resistor is out of the circuit.

JR

Thanks, John.

I might be a little slow on following you here, unfortunately.

Are you simply saying that if I made the attenuator switchable, I'd only have the disadvantage of an "unbridged" connection when attenuation was desired?  I take it there's no good way to have my attenuation and my bridged input, too?

Sorry-- I only have barely enough knowledge to be dangerous, so please bear with me. Thanks again!

soapfoot said:

Thanks, John.

I might be a little slow on following you here, unfortunately.

Are you simply saying that if I made the attenuator switchable, I'd only have the disadvantage of an "unbridged" connection when attenuation was desired?  I take it there's no good way to have my attenuation and my bridged input, too?

Sorry-- I only have barely enough knowledge to be dangerous, so please bear with me. Thanks again!

Click to expand...

And I must be careful in answering you in case you are using the word "bridged" loosely.

My understanding of a "bridged" termination is roughly 10x input impedance to source impedance. Mic preamps target a bridging termination with roughly 2k input impedance for mic's nominal 200 ohm source impedance.

Pad switches routinely maintain, bridged and balanced input behavior in both positions.

Just off the top of my head, a pad switch that alternates between connecting the mic directly to the input transformer when pad is defeated (ASSuming the tx delivers nominal termination), and when the pad is desired dropping resistors of roughly 1k are added in series with each leg from the mic, with a 200 ohm or so shunt resistor feeding the TX.  This way the mic sees similar load in both cases, and tx sees similar source impedance.

There is no need to reinvent this very old wheel. or rely on my butt derived values, look at amy mic preamp schematic with a pad switch for practical real world examples. 

JR

Thanks!

I'm currently using an attenuator built according to this plan: http://community.dripelectronics.com/media/kunena/attachments/69/atten.pdf

The input Tx is the Sowter 9970, which is a 1:7 (200R/10K) mic input transformer. Thinking about it now, even the raw transformer on its own would show a very low load impedance to the microphone, unless I'm understanding it improperly. 

On the attenuator plans above, I may just strip out the pair of 200R resistors on the "-0dB" switch position, thus bypassing the attenuator completely in the clockwise-most position. That should be an improvement over the existing scheme, which (I think) would only show the mic 100 ohms in the "-0dB" position. That would explain why the previous position ("-3dB") is actually louder-- it is (I think) showing nominally 600 ohms to the microphone.

I'd like to devise a way to keep this 12 position attenuator switch (in 3dB steps) while showing the input Tx a relatively-constant 200R while showing the microphone 2k or more.  I'm not sure if there's an easy solution for this.

You want to show 200r to the transformer but 2,000r to the mike.

The lowest-loss passive pad which can do this is a 20dB pad. Not your 3dB teeny-steps.

We can often cheat to 15dB, maybe 10dB depending on mike. But 10dB pad with 200r out is only 620r input. Significant load on some common 300r mikes.

The input transformer may be 200R/10K but the secondary is typically loaded much higher than 10K. The input is typically much higher than 200 ohms. We don't "match" microphones, that throws-away 3dB of precious signal power thus S/N.

PRR said:

You want to show 200r to the transformer but 2,000r to the mike.

The lowest-loss passive pad which can do this is a 20dB pad. Not your 3dB teeny-steps.

Click to expand...

OK. As long as I understand!

The input transformer may be 200R/10K but the secondary is typically loaded much higher than 10K. The input is typically much higher than 200 ohms. We don't "match" microphones, that throws-away 3dB of precious signal power thus S/N.

Click to expand...

Apologies if this is very basic, but: If the secondary is loaded higher than 10k, does it then follow that if we "show" the mic directly to the transformer, the mic will "see" more than the 200R load impedance? In other words, is the 200R primary impedance constant, or does the transformer simply transform down the secondary impedance via the 7:1 ratio?

soapfoot said:

Apologies if this is very basic, but: If the secondary is loaded higher than 10k, does it then follow that if we "show" the mic directly to the transformer, the mic will "see" more than the 200R load impedance? In other words, is the 200R primary impedance constant, or does the transformer simply transform down the secondary impedance via the 7:1 ratio?

Click to expand...

Impedance transform is turns ratio squared so 49:1.

I don't follow your switch topology.

As PRR shared the standard pad topology limits the range of attenuation while maintaining optimal source and input Z.

You can probably cheat one or the other end.. Driving the mic from less that 200 ohms probably won't cause too much damage to signal integrity. (IMO) Driving the transformer from higher Z could cause response issues.

Good luck...

JR

Ah!  Thank you all very much for bearing with me. Makes sense.

I haven't noticed anything sounding "wrong," so maybe it's all academic, anyhow.

> If the secondary is loaded higher than

Transformer is a lever.

It doesn't have "an impedance". Or rather what it has is two (and more!) impedances where it starts to work bad.

I have a modified nailfile for prying chips out of sockets. I have a 3-foot prybar. One will lift 0.1"-0.2", the other 2"-6". While each has a range, I would not pull large spikes with the chip-pry, I would not pull chips with the big prybar.

So the "200r" on the tranny is like the "4 inch pull" on the prybar. You don't have to hit the exact spec, but if you get far off you will be sorry.

The "200r" winding has about 20 ohms of copper resistance. If we use it in a 100 ohm circuit, it works, with perhaps 20% loss. If we work it in a 20 ohm circuit, half the precious signal is wasted. OTOH it has a shunt impedance perhaps 2K (more likely 1K at the band limits and >10K midrange). If we used it in a 1K circuit half the power is wasted (perhaps worse: big loss at 20Kz and 20KHz but low loss at 1K, a humpy response).

So your "200r" should be fed from a 100r-400r source. This covers most mikes. (The odd 50r powered condenser mike is so hot that loss or noise-resistance is unimportant.) (This will not be the best choice for old-old RCA mikes wired 37 ohms.)

The voltage ratio is 1:7. The current ratio is 7:1. Impedance is voltage over current. The impedance ratio is thus 7*7 or 50. (If your abacus gives 49, it's too fussy.)

50*200r is 10K secondary.

We don't load the secondary much. Probably higher than 100K, though maybe less at 20KHz (zobel). 10K reflects back as 10K/50= 2K presented to the microphone.

This is all Voltage Transfer. Same as your house. You have 120V no-load. To get the absolute maximum amount of Power in your house, you would "match" the power line losses, increase current, pull the voltage down to 60V. That's the most power you can get in your house. But there is an equal amount of power heating the power line outside. Somebody pays for that wasted power. Also there is the problem of 120V lamps going dim at 60V or 60V lamps popping if you cut other loads and the line un-sags to 120V. So power systems are designed for "low" losses. Several layers of 2% loss (2% from dam to street, 2% in your feeder, 2% from fusebox to outlet).

Note also: a power systejm has "an impedance" but it isn't directly considered. My 240V house can suck 2 Amps at 3AM or 40 Amps on a cold hungry evening. 240V/2A= 120 ohms, 240V/40A= 6 ohms. My power line has an exceptionally high 0.4 ohms impedance. We can see why my lights dim when the radar-range starts.

Because audio power is not "real power", we don't throw tons of coal or water over the dam to make a milliwatt of audio, the economics and "optimum" losses are different. Losses may be 5% to 20% depending on other factors.

Thanks, that's really informative. I'll read it a few more times to make sure I'm understanding it better.

A few questions I have, if you don't mind taking the time:

PRR said:

The "200r" winding has about 20 ohms of copper resistance. If we use it in a 100 ohm circuit, it works, with perhaps 20% loss. If we work it in a 20 ohm circuit, half the precious signal is wasted. OTOH it has a shunt impedance perhaps 2K (more likely 1K at the band limits and >10K midrange). If we used it in a 1K circuit half the power is wasted (perhaps worse: big loss at 20Kz and 20KHz but low loss at 1K, a humpy response).

Click to expand...

OK, so to make sure I'm understanding right: the primary winding has about 20 ohms of resistance-- this much is fixed and is always a component of the impedance "seen" by the microphone, correct? 

Due to Ohm's Law, the impedance ratio (from primary to secondary) is 1:49-- the transformer's voltage ratio is 1:7, and its current ratio is 7:1.  Impedance equals voltage divided by current.  7÷(1÷7) = 49. So the impedance seen on the primary side of the input Tx would be [(impedance on secondary) ÷ 49]+20.  Am I on the right track?

I'm not sure yet how to figure out the input impedance of an EF86, but if I had that number, I could figure out the actual impedance reflected back to the mic, correct? And if that number is higher than about 10k, the actual impedance seen by the mic will be more than 200R.  Correct?  And this will vary somewhat with frequency too, right?

One thing that I'm not sure I understand yet is where the 2k "shunt impedance" number comes from. I'm googling, and learning, so I'll keep working on it. Compared to most people here, my understanding of certain things is still very basic.

So your "200r" should be fed from a 100r-400r source. This covers most mikes. [size=8pt](The odd 50r powered condenser mike is so hot that loss or noise-resistance is unimportant.) (This will not be the best choice for old-old RCA mikes wired 37 ohms.)

Click to expand...

I think I'm a little lost by this point.  Wouldn't a 400r source want to see about 4k input impedance for maximum voltage transfer?  Is the "200r" transformer winding actually showing it that much?  I just want to make sure I understand.  This would imply that the secondary of the input transformer was seeing about 200k from the EF86.  At 100k or above, it would at least give you a solid 5x the 400R source impedance, which is pretty good.  Am I missing it?

Thanks again for your time and insight so far.

soapfoot said:

One thing that I'm not sure I understand yet is where the 2k "shunt impedance" number comes from. I'm googling, and learning, so I'll keep working on it. Compared to most people here, my understanding of certain things is still very basic.

Click to expand...

The 2k termination target comes from desired bridging termination (10x) for good voltage transfer.

2k load comes from roughly 100k termination across secondary of Tx.


JR

Got it, makes sense.

So with the transformer alone (no attenuator between), the input Z of the preamp would be somewhere around 2k, most likely. Right?

hopefully...

JR

Yeah, I think probably so. Just an EF86 grid on the other side of that transformer.

> Wouldn't a 400r source want to see about 4k input impedance for maximum voltage transfer?

It isn't gold dust. We don't need EVERY last flake of sound. If you are styanding there with a 400 ohm mike, and a 200r nominal 2K actual mike input, plug and play, rock and roll.

> where the 2k "shunt impedance" number comes from.

Curveball; and even JR missed.

There's eddy currents in the iron. Iron resistance is high but not infinite, even when we slice and oxidize. Looking into an unloaded transformer, we see a resistance in shunt (parallel), which sets some upper limit on practical impedance.

Copper losses set a lower limit on useful impedance.

(Additional losses at frequency extremes.)

The art of transformer design is to pick these losses to "all be tolerably low". This requires some notion of a Nominal Impedance. Often we can get each loss well under 5%. That's good enough.

(In Utility Power transformer design, full-load losses are held under 2% and no-load losses super-low; power is money and transformer losses are before the cusomer meter!)

If you go a little off the nominal impedance, some losses increase and others decrease. You can measure the difference, but little harm is done to the audio.

Sometimes you can cheat. The $2 transistor-radio transformers are rated 300Hz. If you drive a "10K" winding with a 100 ohm source, the ~~900r DC resistance allows small bass to pass down to 1/11th of rating, below 30Hz! (It still may not take high-level bass; you can only force things so far.)

Build yourself an impedance meter. Signal generator, resistor-box in series with tranny winding, audio voltmeter. Pick your level to suit the iron: 1V for big line cores but maybe 10mV for small mike transformers. Diddle resistor. When voltage across winding is half of signal voltage, impedances are about equal (let's ignore the imaginary vector). You will find that a good 200r:grid tranny is above 200 ohms at 20Hz, above 2K over most of the band, can be over 10K midband, but then drops because "open grid"  (and transformer) is significant capacitance.

PRR said:

> where the 2k "shunt impedance" number comes from.

Curveball; and even JR missed.

Click to expand...

JR defers to PRR concerning esoteric transformer factoids. I learned decades ago I can do better and cheaper without them so stopped trying.

FWIW I did try combining cheap transformers with (relatively) cheap opamps (back in the '70s) to try to get opamp like performance from cheap iron. That exercise ended without success. 

JR

When you start needing to know this level of detail, you need to know that the impedance varies with frequency, and at various extremes it will be very far off from nominal.  This affects phase and frequency response, including frequency response of the mic itself in some cases.  Don't freak out, just rock and roll.  Simple answer as always is the basic 20 dB U pad, anything else requires lots more knowledge and lots more second-guess worry in use, ignore the internet noise about 'mic pads mucking up the sound' and get on with it, unless you want to go down the endless rabbit hole for every mic and preamp combo in your kit.  You could sort out 'optimal' for every combo, and you'd also never get anything done. 

This is why I come to GroupDIY.    Years later, the discussions and great answers find their way forward.