Looking for a transparent solution for a pre-amp

[allances]

Peranders,

I am building a multi-channel pre-amp for my system, and the initial block is a passive attenuator based on a ladder-type, relay-based pcb from twistedpearaudio.com .(Joshua Tree)

My concern is that my interconect cables are long(all the five speaker have their own amplification close to the baffle, Linkwitz Orion's with DIY amplifier using UCD modules), specially for the surround speakers, so even though the attenuator is of the lower type impedance(2k ohms in and 2kohms out), I would prefer to have an even lower output impedance(the crossover input impedance is 20K, if I am not mistaken...).  I was considering using dual-op amps to shield the attenuator from impedance concerns regarding the source output and the crossover input and long cables, but I believe a discrete buffer could be so much more transparent. I believe the super-buffer is a possible solution, given the high-performance, the class-a configuration that can be used, but I would like some advice on how to use it, and even if it wouldn't be better to use the qrv-08 project, given the super-regulated power source is already included on that project...

One thing that worries me is that, unless I am able to get a very good source, I might have to include a buffer stage BEFORE the attenuator too, since the 2K ohms load might be too high for some high impedance outputs that are most common with mid-end gear. Is that reasonable, or I am being over-cautious without reason? I don't want to add anything to the system that is not needed to extract best performace with minimal transparency sacrifice.(although some might argue that after 5-8 more op-amps on the crossover, i shouldn't care for not using op-amps on the pre-amp...)

Thanks,

Allan

[peranders]

All of my headphone amps are excellent as buffers but possibly a bit overkill.

You should feed a long cable with a low source impedance and by that minimize the effects of the cable capacitance.

So, the attenuater first and the the buffer.

[allances]

Peranders,

I finally completed building my pre-amp. I've used the topology:

signal in -> buffer -> attenuator -> buffer -> signal out.

I am feeding all the diamond buffers from a single +15V - 0 - -15V PSU but i do have problems...

First, when the input signal is open, i can measure 15V at the input of the first buffer. I did expect R1, R2 and R3 to keep that much closer to 0.

The second problem is probably related to the first. When the attenuator is connected and the input signal is set to 0v, i can measure 6mv at the attenuator input.(I expected that since my first measurement with 0v on the input of the buffer gave me an output of 4-6mv on all tested buffers), but i measure 1 volt at the attenuator output(connected to the input to the second buffer stage, the voltage is coming from the second buffer stage) and get 1 volt at the second buffer output.
My theory is that since the attenuator gives a low voltage signal, but with high impedance(2R-R attenuator, never connects the output directly to the ground, always have a resistance between output and ground), the second buffer is seeing an input that is midway between 0v and an open terminal(infinite impedance). With the tendency to present 15 volts at the input when it is open, it just gets a value that's between 0v and 15v(1v in this case, and it is consistent among two different output buffers)

Now, the first problem seems to be the key: How do I make sure the input signal stays at 0v when the input is open? Should I remove R2 and R3? I suppose R1 gives me the 1Mohm input impedance of the circuit, so I do not want to remove that, right?

Thanks,
Allan

[peranders]

The diamond buffer must always have DC bias at the input so if it works alone you must either have it AC coupled and trim the base current. The other solution is to have a low ohmish signal source.

[allances]

Ok, the buffer will not be working alone. It is always connected to something. What I am trying to do is to be sure that the second stage gets 0v at its input even when connected to a high impedance source.(2k-20k) Since i do not understand how to do DC bias, I really need a more through explanation of what I need to do. I think i need to be sure something in the circuit keeps the input at 0 volts by feeding the necessary current to the adjacent transistors. But this something cannot present a low input impedance to the source...

I am not sure the PM worked, so thats why I am repeating it on a forum post...

Can you explain me how R1, R2 and R3 work on this circuit, and why would i not use some of them(or all of them...). This is briefly mentioned on the build instructions. Also, I am using the class-a circuitry.

Allan

[peranders]

The input of the diamond buffer has a base current which can be either positive of negative. This current must always be able to flow. How much it is in your case can you simply measure by shorting the input and measure the voltage across R5. If you have this buffer in combination with an opamp this isn't anything you have to be bother with but if it's used stand alone it is.

T13 and T14 should not be used at all in your case. This circuit is to draw a constant current from an opamp output (which you don't have here).

R1, R2 and R3 is only place holders for resistors. In your case you can have R1 and the value from 100 kohm to 470 kohm or maybe more. R1 is to prevent a floating input which is not good. It may create any output voltage from plus to minus, quite nasty.

[allances]

Peranders,

Now I am understanding it... Since i have no opamp(the buffer is not being a current source for a higher impedance opamp output, it is being used as a high impedance voltage mirror/current source to a passive attenuator), I should not use T13 and T14 since they need to feed from the current of said opamp. This also explains why the first buffer works fine when either connected to the source or when shorted to ground.(both cases are low impedance)

I got R1, to keep it from having a floating input. Not sure i understand how R2 and R3 are used though... in my case, i should use just R1(1Mohms or less, depending on what i want for input impedance) and remove R2 and R3?(this actually makes more sense to me than having all three...)

Regards,
Allan

[peranders]

R2, and R3 are used if you don't have a dual power supply or for some other reason.

[allances]

Peranders,

I removed the transistors, and things are looking much better. I will remove R2 and R3 since I am using dual power source, but without removing them(so now i have R1, R2 and R3 as 1Mohms resistors), and with the input shorted, i have 0mv around R5. and 8mv at the output. Just removing the transistors was enough to let R1 do it's job. Now, with input open, i got about 150mv on the input(it seems to vary amongst different buffers, some have 30mv, others 80mv...) I think the differences in inputs between these buffers is related to the voltage divider currently formed by R2 and R3.(which fortunately should float around 0v...)

I tried measuring voltage around R5 with the input shorted, but I swear I've got 0v.(one of the four actually have 0.2mv)

Allan

[peranders]

0.2 mV across 560 ohms is a quite small base current and some of the modules has less than 0.1 mV. This means that the current gain is rather alike between your PNP and NPN transistors.