Joshua Tree Attenuator, suitable input impedance for the 3-channel B22?
Jul 26, 2010 at 1:37 AM Thread Starter Post #1 of 71

Lil' Knight

Headphoneus Supremus
Joined
Sep 16, 2007
Posts
6,872
Likes
23
I've just received the JTA kit from TPA. According to their website, input impedance varies between 2.2K and 10K. My source has the tube output stage which has ~1K output impedance.
 
My question is what input impedance I should use for this setup?
 
To my understanding, the amp's input impedance should be 10x the source's output impedance, so I guess the 10k should be fine. Not sure if there's a manual for the JTA somewhere but I don't know if I could get a fixed 10K input impedance.
 
Any opinion regarding this is appreciated.
 
Jul 26, 2010 at 7:46 AM Post #2 of 71
You can't get a constant input impedance for the JT. It sacrifices that for a constant output impedance instead.
 
I'm pretty sure that Russ has previously offered resistor kits for higher input impedances, but this also raises the output impedance. I don't think that the B22 would care though.
 
Jul 26, 2010 at 9:50 AM Post #4 of 71


Quote:
Originally Posted by Lil' Knight /img/forum/go_quote.gif
 
Would it be possible to use the Delta-1 calculator for the JTA? http://www.amb.org/audio/delta1/r2r.cgi
 
What are the negative effects if the JTA's input impedance is too low?



No idea on the Delta-1. But I doubt that the architectures are exactly the same, so probably not.
 
Negative affects are that your tube output stage is working much harder than it needs to. But I'm not fully versed in such things......
 
Jul 26, 2010 at 10:31 AM Post #5 of 71
WHY!!!???
 
The whole point of the J-tree is that it has a constant and lowish output impedance which is necessary for certain power amps that have either highly capacitive inputs (simple FET amps with no global feedback, and the like) or low input impedances (such as may be found when using an inverting opamp or in amps with a low input impedance for other reasons). In any case, it is a specialized piece of gear. For an amp like the β22 it is good for nothing but fapping.
 
Get a normal potentiometer. Take every penny you would have spent on the J-tree and buy the most expensive 10 or 50Kohm pot you can find.
 
If you must have a piece of gear that nobody understands how it works: get the broskie/tubecad stepper. Just like the J-tree stepper there are very few people who understand how the Broskie stepper works so you have still obtained mystery. Its a bit of a PITA with 3 knobs, but mneh, its a small price to pay. You could always read about how the Broskie stepper works and design your own with different switches requiring only 2 knobs (coarse and fine)
 
Jul 26, 2010 at 11:08 AM Post #7 of 71

Yup, it has an input impedance that varies down to 2.2k as you use it, and hence is little or no use at the input of a normal headphone amp. Load your 1k tube source with 2.2k and it will most likely heavily distort.
 
Note that most attenuators have a constant (or at least reasonably high) input impedance and a variable output impedance, not the opposite as with the JT.
 
If you need a relay stepped attenuator the Tent Labs unit may be suitable as it has an input Z always above 40k ( http://www.tentlabs.com/Components/page31/page31.html ) .
 
Jul 26, 2010 at 11:25 AM Post #8 of 71


Quote:
Yup, it has an input impedance that varies down to 2.2k as you use it, and hence is little or no use at the input of a normal headphone amp.



I don't necessarily agree with that. 50k seems to have become the default standard for solid state amps, but for no particularly good reason. A lower input impedance does certainly have benefits, and a constant output impedance can be very desirable.
 
The only qualification is that your source has to be able to handle it. Most solid state sources have output impedance well under 100 ohms, and can easily drive the 2.2-10k of the JT.
 
Jul 26, 2010 at 11:47 AM Post #10 of 71


 
Quote:
 
The only qualification is that your source has to be able to handle it. Most solid state sources have output impedance well under 100 ohms, and can easily drive the 2.2-10k of the JT.


Yes, in that instance I was speaking in general terms in respect of the input impedance. I am sure you are right that some sources can drive 2k2. Indeed, as you intimate, a low source impedance will not loose much signal level into such a load by potential divider action. However the issue with driving low impedances is often not loss of level, but simple distortion. Even with output impedances of 100 Ohms some sources will not like 2k2 loads, and distortion will result. I am sure you are aware that most op-amps will happily drive 2k2, however most cathode followers (tube equipment) and simple low-current emitter followers (older solid state equipment) are likely to be unhappy with 2k2.
 
So I agree that it's a case of working out if the source can handle the worst-case load. If we DIY that should be OK, but I would be reluctant to purchase an amp that went down to 2k2 at the input because of simple concern over potential matching to any future source that may be specified into 10k or greater, as most seem to be.
 
Jul 26, 2010 at 11:59 AM Post #11 of 71
 
Quote:
I don't necessarily agree with that. 50k seems to have become the default standard for solid state amps, but for no particularly good reason. A lower input impedance does certainly have benefits, and a constant output impedance can be very desirable.
 


Constant input impedance only matters if the output of your source is dependent on load impedance being constant. I guess we are talking about tube output stages, its best not to take chances with something that varies from 10 to 20Khoms. [mockery]Who knows if it can handle that.[/mockery]
 
Quote:
A low input impedance surely has less noise. I think 20-25k is perfect for my setup. Still waiting for the reply from TPA regarding raising the input impedance.


It can be done. But WHY?
 
The whole point of the J-tree is that the OUTPUT impedance is low and the same at all levels. Not only dont you need that at all, but you are going to screw up the output impedance by making the thing work the way you want. So you have taken something that dosnt work right for your application and made it work worse for every application!
 
I am pretty sure that by increasing the input impedance of the J-tree you increase the output impedance significantly... perhaps to the point where you would be better off with a 10K or even 25K ohm pot.
 
I understand that TPA is in the lexicon of approved head-fi kit suppliers, but believe it or not there are others. Put the $150 towards half of a P+G pot, Or just get a DACT (whose steppers are in the lexicon of head-fi approved components), Or get a TKD and spend the leftover $50 on beer.
 
Jul 26, 2010 at 12:52 PM Post #13 of 71
Another option is to use a different volume control.
 
Do you mean use a buffer in place of an actual gain stage? An "amp" with gain of 1? This is a great idea, but not supported by head-fi. Sorry.
 

Users who are viewing this thread

Back
Top