impedence balance a PM1000

[buschfsu]

anyone try this. i don't have the cash for the tamuras and id like to interface with balanaced gear across the room.

electrically balanced out isn't for me as i don't want ic's in my discrete circuits@@!!

 

[Greg]

I'm not sure what you're asking... can you be a little more detailed.

I think you're saying you can't afford a transformer and you don't want to use ICs to balance?

 

[JohnRoberts]

[quote author="buschfsu"]anyone try this. i don't have the cash for the tamuras and id like to interface with balanaced gear across the room.

electrically balanced out isn't for me as i don't want ic's in my discrete circuits@@!![/quote]

Impedance balance is mostly a "it couldn't hurt" and since it only only cost a few pennies for the resistor, if you have the time and desire, go for it. To do no harm it needs to connect to the clean signal ground reference at the sending unit.

There won't be much practical benefit from the impedance balance over short distances, but proper input differentials are always useful at any distance.

JR

 

[buschfsu]

so as far as matching goes i need to put a resistor into the circuit where the output goes to ground. that part i get, How do i calculate the proper impedance value that will 'match' the unbalanced output?

thanks

 

[JohnRoberts]

[quote author="buschfsu"]so as far as matching goes i need to put a resistor into the circuit where the output goes to ground. that part i get, How do i calculate the proper impedance value that will 'match' the unbalanced output?

thanks[/quote]

Indeed as Roger said, popular practice is to use same value as build out or source impedance of single ended output. Note: if your receiver isn't true balanced the impedance matching is less important. It may just serve to open a ground loop between poorly designed gear.

JR

 

[buschfsu]

heres the schematic, i took the output for the unbalanced leg after the cap c37 in the highlighted area.

there isn't one serial resistor before this cap so im alittle lost.

thanks

http://bellsouthpwp2.net/b/u/buschjason/pm1k schem copy.gif

 

[JohnRoberts]

[quote author="buschfsu"]heres the schematic, i took the output for the unbalanced leg after the cap c37 in the highlighted area.

there isn't one serial resistor before this cap so im alittle lost.

thanks

http://bellsouthpwp2.net/b/u/buschjason/pm1k schem copy.gif[/quote]

Your source impedance there is 11 ohms reduced by loop gain margin of active circuit (pretty low at audio freq). You probably need to add your own build out resistor (100-300 ohms) to isolate non resistive loads. I'd use same value for both.

JR

 

[Guest]

Use couple of 300 Ohm resistors, one from output (hot out), another from ground near the output (cold out).

 

[buschfsu]

id love to know how you figured 300ohms (teach a man to fish you know)
are you trying to let the next stage see 600ohms?

 

[buschfsu]

if you are looking at the circuit i posted its the yamaha pm1000 discrete channel strip from the 70's. Lot of folks really like it.

 

[NewYorkDave]

Add enough resistance to the high side of the output to give stability and protection when driving reactive or low-Z loads. Then add an equal resistance in the low side. 300 ohms per leg is probably excessive for this particular circuit; you'd lose 6dB across those buildout resistors when driving a 600-ohm load.

 

[JohnRoberts]

[quote author="mediatechnology"]Yes, that's the point. Use 301 ohm 1% at the output "tip," and in series with the ground to produce your "ring" connection. Sum is 602 ohms.

They probably don't need to be 1% but 1% are cheap.

JohnR: Is that a bootstrap at the base of Q3? (And also at Q1?)

Interesting pre-IC circuit BTW. Virtually textbook. Like the sound of it?[/quote]

Sorry, I didn't see this until just now.

I'm not sure the C at that circuit point is doing much since it's in parallel with a diode which is pretty low impedance. Perhaps that's why it's a 100uF for any effect. It will perhaps reduce crossover distortion somewhat at HF.

I agree with NYD regarding build out impedance... there are sundry philosophies, but like philosophy there are different schools of thought.

The matched 600 ohm driving 600 ohm is optimal for maximum power transfer but archaic and obsolete in today's world of "bridging**" termination for maximum voltage transfer.

Another popular philosophy is to match the characteristic impedance of the wire used (50-100 ohm). While this might reduce reflections and improve signal integrity in theory, in practice the length of audio wavelengths relative to lengths of cabling makes this moot also.

Finally, it is of value to decouple capacitive loading, such as shielded wire and perhaps crude input filtering on following gear. In my experience 100-150 ohms seems more than adequate for that using modern opamps. I have seen consumer or nonprofessional gear using opamps of limited drive capability to use build outs of 1k or larger. This is no doubt based on the premise that it's better to drop some signal level and play cleanly limited by voltage clipping into PS rails than to current limit at lower levels.

As there is no single best answer here, I'm inclined to look for the range of wrong answers. I would target a bridging interface termination so source impedance needs to be <1/10 of anticipated input impedance of following gear. While decades ago I designed gear that would do this even into a 600 ohm input, I can't remember the last time I saw a 600 ohm input on a product made in the last few decades. So nominal input impedance is more like 10k giving us a 1K upper limit. Another consideration wrt this upper limit is the LPF formed by this R and cable capacitance (30 pF/foot).

So we have an upper limit of 500 ohm per leg. The lower limit is unknown since this design was not intended to drive outside the box we have no idea of it's stability.

Depending on how seriously you take all this, I might suggest experimenting with range of small build out resistors and few nF to ground to simulate nasty load. Either look at a square wave response with a scope for ringing, or just look for oscillation. when you determine how much R you need for stability, double that for a safety margin.


JR


** Bridging in this case refers to an input termination that is 10x the source impedance for insignificant degradation of S/N and maximum information transfer.

 

[NewYorkDave]

Yes, rumors of the demise of the 600-ohm load have been greatly exaggerated :wink:

In most cases, I design the outputs of my gear to be capable of driving 600 ohms to at least a reasonable level without excessive distortion. Of course, better results are usually obtained with a higher impedance, "bridging" load--although when using transformer-coupled gear with a nominal 600-ohm secondary, you may have to load it with 600 ohms for flattest frequency response.

As far as output impedance goes, lower is usually better. The cases in which you actually want a 600-ohm output impedance are relatively rare, the notable exception being passive EQs or filters that depend on specific values of source and load impedances. Even then, it's easy enough to add buildout resistors to low-impedance outputs when you need to satisfy that condition.

I agree with John that, at the cable lengths we usually deal with in studios, at audio frequencies, characteristic impedance of the transmission lines is usually something we don't have to try to match. I work in broadcast, so I do encounter those cases from time to time when cable lengths are significant with regard to signal wavelengths (e.g., STL over landlines of a mile or more in length).

 

[JohnRoberts]

Getting back to the OP's question, I wouldn't automatically ASSume that circuit to be happy interfacing with 600 ohm inputs, so perhaps test for that too.

Agreed if planning to interface with antique gear use antique interface standards... :idea:

JR