diyAudio (
-   Analog Line Level (
-   -   Building a Tee attenuator (

dimkasta 5th May 2012 11:51 AM

Building a Tee attenuator
I am designing a 0-45db stepped Tee attenuator for my rig.
I have chosen to do this over my existing L-pad because of various advantages.

My issue is how to properly select input and output impedance.

My sources have 27R and 600R impedances respectively.
My amp has 330K impedance.

I currently use a lightspeed attenuator which basically works as an 8k L-pad attenuator.

What I was thinking is to use at first a 9K - 50R combination which will allow me to go from 45db up to 28db.
And then increase the ouput impedance slowly and incrimentally to allow for lower attenuation.
100R will allow ~25db
200R for ~22db
400 for ~19db
800 for ~16db
1600 for ~13db (my usual top volume)
3200 for ~10db
and so on

Any ideas/suggestions?

DF96 5th May 2012 01:55 PM

Why? T pads are used when you want to maintain particular in and out impedances e.g. for RF. In audio the requirement is different (and simpler): as high as possible at the input (so you don't load the source), as low as possible at the output (so you can drive the load and any cable). An L-pad does this, which is why L-pads are used.

dimkasta 5th May 2012 04:06 PM

Well the tee is a design decision that serves and solves other design issues in my pre.

And about the L pad, it controls only the input impedance. The output is low on low volumes but varies widely as the volume increases and can reach easily 1,5k or more

jameshillj 6th May 2012 03:01 AM

Evening Dimitri,
Your system sounds pretty well thought out with a nicely high amp 330kR and the sources 27R and not so nice 600R - if the 10:1 rule still applies, input Z of the attenuator should be above 6kR so the 9kR design point looks pretty okay.

If the amps input Z is 330kR, then your increasing output reistance setting would come out to something like 33kR for a possible zero attenuation hich is still within that 10:1 rule of thumb ratio - This has a strange "feel" about it but I can't give any reason for it.

If you were to initially choose the same 9kR value (and reduce it progressively later), this would a simpler start and also keep the 2 series LDRs approx equal value - don't think you need to go down to anything like 50R o/p Z - just my opinion, no facts, and it's been some years since I last looked T, Pi and H pads.

Sometimes these impedance "rules" are totally U/S - my cd player's has 2 o/p stages (low Z of the 627 IC and 200R of the DCB1), but the amp's inputZ is only 9kR (the FirstWatt F3) - the standard lightspeed gives a reasonable sound but not so clean on the bass with the low source impedance - the suprising thing is that this doesn't deteriorate much with the higher 200R source resistance of the DCB1 - not sure why.

The strangest thing is that Alan Flores' unit (the Warpspeed, supposed Z of 5kR) also uses the L-pad configuration but is completely untroubled by the low amplifier impedance of 9kR amplifier - the low bass clarity and accuracy is a real surprise and is the most transparent vol control I've heard without any extra treble "peak" or distortion - no idea why this is so, but very happy about it, naturally!

(naturally, it would be better to add a buffer in front of the F3, and will possibly be required when I add the Aikido gainstage in the system, but who knows .... and a T-pad vol control might make it unnecessary.)

I'm playing around with Uriah's LighterNote and it'll be interesting if I can get the o/p Z down to about 1kR without totally stuffing up the source loading - one of the guys uses it successfully infront of his amplifier that has an input load of less than 4kR and it works perfectly, so suprising things about these 'so-called' simple lightspeed gadjets!

This is the first time I've seen a serious attempt at something other than the basic L-pad - congratulations. You've just become a 'guru'!

dimkasta 6th May 2012 12:52 PM

guru... that s a lol :)

The biggest issue with the tee so far is the increased accuracy required. In the range of 1/10th of an Ohm.
Especially since I plan on doing it with LDRs where there is easily some fluncuation of 2-3ohms / 1-2 secs either due to temp changes, or due to psu instabilities.

I think I am losing interest quickly...
I am back looking at buffers...

jameshillj 6th May 2012 01:56 PM

A series voltage reg and a current pump takes care of the power supply variations (like Uriah's system) and physically coupling the ldrs together eleviates the temp drift, or keeps it same for all devices.

I'm a bit puzzled as to what element of the T requires 100mR accuracy? As the network increases in sophistication, the individual element accuracy actually reduces as the whole network absorbs much of the variation - or, in other words, with the same matching of the devices in a T network instead of an L-pad, the relative accuray or channel impedances will be much closer.

Buffers can introduce their own signature, unfortunately and the most common one is signal compression, just what we're trying to get away from with standard vol pots. The B1 or the complementary version (k170/j74) work pretty well but do require good supplies.

... my 2c.

AndrewT 6th May 2012 02:10 PM

I too am lost on the logic that claims that a T attenuator is better for audio volume control.

A T attenuator would be great if one has a 75ohms source feeding a 75ohms cable into an attenuator, which then drives another 75ohms cable feeding a 75ohms receiver.
If HF is being handled then the matching of impedances at each handover is important to passing an undistorted HF signal.

Frank Berry 6th May 2012 02:15 PM

Daven made "T" audio attenuators many years ago. They were used in vacuum tube mixing consoles where 600 ohm input sources needed to match the mixing buss which was transformer-coupled to the Program line amplifier.
This was back in the days when all "broadcast" gear used input and output transformers.

AndrewT 6th May 2012 02:23 PM


Originally Posted by dimkasta (
......... The output is low on low volumes but varies widely as the volume increases and can reach easily 1,5k or more

Yes, that applies to all unbuffered volume controls.

This is easily solved by adding a buffer such that the Source (vol pot) can drive the cable, or by eliminating the cable and attaching the vol pot directly to the Receiver circuits, i.e. inside the Receiver and preferably on the PCB right next to the input device/s.

The output impedance itself is not the demon, It is the interaction of output impedance and parasitic/deliberate capacitances that can become the demon.

at and near maximum volume, the output impedance is equal to, or very nearly equal to, the Source impedance.

All times are GMT. The time now is 05:27 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio