THAT preamp experiments......

Thanks for the schematic.

It is impossible to predict what that cap will do without looking at the specific input circuit it is connected to. Further it will very likely act differently when connected to different destination inputs.

The fact that you report hearing a difference suggests there is signal current flowing back into or out of pin 3 from the destination termination. This voltage is subtracted by differential action at the destination.

I would be inclined to add a small R (50-100 ohm) between pin 2 and the THAT chip output line. Likewise a similar value R from pin 3 to ground.

If you wish to add a high pass filter to the signal perhaps locating it in series with the very inputs would give you the most predictable and stable results, but these would probably be electrolytic for typical cutoff values and perhaps not the most ideal capacitor dielectric.

A C in series with RG would likewise add a HPF pole in the gain function, but again would require a larger electrolytic value.

JR

I've been running a pair of 1512 pres on +/18V for a year or so without issue.

EDIT: Now that I realized there is a page 2, I see that this was already addressed.

Another 3AM night at the studio last night. So tired.

BTW, someone on the micbuilders-list made straight ahead PCBs for the 151x (no frills).... He had to order a few more than he needed himself, but dunno how many he has available.
So if you're the kind of guy that needs an adapter-PCB for TX-rewiring you might be interested :wink: :twisted:
Ha, being lazy these days I'm getting a few myself as well :green: , those 1510-samples from Wayne should see some action.



Regards,

Peter

I'm not trying to be a wiseass, I'm just checking what I think I've learned about impedance balanced outputs. So correct me if I'm wrong.

- we need to know the output impedance of the circuit we're dealing with.

- we then add a couple of dozen ohms in series to avoid oscillation.

- the undriven output (Pin 3) is then created by a resistor that mirrors the driven output's impedance i.e. circuit output impedance + couple of dozen ohms.

- in most cases (but maybe not here) a large lytic is required to block DC offsets. A same size cap must be used in each line, driven and undriven.


Speaking of THAT chips, wouldn't one of those THAT output chips be a nice solution for a transformerless, clean and easy balanced solution?

Yup.. for optimal CMR in balanced receivers the + and - source impedance should be identical, so external noise will be picked up identically by both lines.

Also adding the resistor to a single ended output allows the marketing types to call it an impedance "balanced" output. A lot of perceived value at point of sale for 1/2 cent in parts cost.

JR

I changed the cap...i put it in the line out to xlr 2 and it changed a lot!
doing that with another cap in XLR pin 3 it affects too.

I don't know how to explain this technically i don;t know if it makes sense..

A C in series with RG would likewise add a HPF pole in the gain function

Click to expand...

john maybe all the changes are just a HPF reactions...you know from lows to thick sound....?

[quote author="Rossi"]I'm not trying to be a wiseass, I'm just checking what I think I've learned about impedance balanced outputs. So correct me if I'm wrong.

- we need to know the output impedance of the circuit we're dealing with.

- we then add a couple of dozen ohms in series to avoid oscillation.

- the undriven output (Pin 3) is then created by a resistor that mirrors the driven output's impedance i.e. circuit output impedance + couple of dozen ohms.[/quote]
Same here, that's how I look at is as well.
In principle one should have a look at a few more frequencies for the output-impedance of the active part: depending on the circuit-topology & dimensioning the output can/will change for different frequencies (let's for now say it usually increases for higher frequencies, among others because of the 'vanishing' influence of neg. feedback).

Stating the obvious in a likewise manner as you did above (so this is how I see it, please shoot):
for good suppression of the nasty stuff, the frequency/spectrum of the wanted signal is irrelevant; what matters is that the incoming disturbing frequencies arrive at identical impedances and hence develop identical signals on the lines (the lines that are subtracted from each other later on at the receiver).
Since there's no guarantee that the incoming nasty stuff is 'like audio'. There'll be hum of course, but at the other end of the audio-spectrum and beyond that there can & will be a lot as well.

So if we want to get a bit more deluxe (or in case we still don't fancy making music with our DIY-stuff) then we might think of mimicing that increasing output-impedance a bit more closely for the 'dummy-side'.

I can't recall having seen this in such circuits though, possibly the figures don't add up to spend additional components here for most situations. I bet such circuits are out there though, so if you know one please post.

- in most cases (but maybe not here) a large lytic is required to block DC offsets. A same size cap must be used in each line, driven and undriven.

Click to expand...

True.
And if we still hate to start making music :wink: we might contemplate biasing the dummy-side electrolytic in a comparable fashion as the one at the other side.

Best regards,

Peter

The source impedance of an active device will change slightly due to falling loop gain margin at higher frequencies but these changes will be modest compared to say a 100 ohm build out resistance in series with that output. To force balance at very high frequency, matched shunt caps to ground at both outputs could dominate there, but for very HF CMR may be compromised by a number of factors.

Mitigating against RF IMO requires a mix of CMR and LPF at receiver. Output stages should be fast enough to not rectify any RF that sneaks into their feedback path via the output. Here the build out resistance divides down RF reducing it's capability to cause artifacts in output stage electronics.

JR

What really surprised me was that Bill Whitlock of Jensen Transformers, in the AES lecture somebody posted a few months ago, said that impedance balanced outputs are a perfectly okay solution. IIRC he liked impedance balanced better than most chip based balancing solutions, excluding, of course, his own "in genious" topology (which THAT uses).

IMO the bulk of the benefit from balanced I/O interfaces comes from the differential math performed in the receiver, which mostly keeps signal grounds properly referenced, forward and back. This differential math will also subtract out any noise induced equally into both lines.

Equal value, opposite polarity signals routed within a twisted pair will generate less crosstalk into other nearby cables but that is generally not an issue with line level signals, run inside shielded cables.

Finally the outputs from most products using opamps or otherwise power supply limited to circa +/- 15V will pick up 2x the signal swing using dual driven outputs. In theory you could get 6 db more S/N, but in practice you run similar nominal signal levels, so the differential output just buys you more headroom, ASSuming the following receiver can handle the 6dB hotter maximum level. If not it buys you nothing but reduced crosstalk, that wasn't rally a problem in the first place.

In addition to being dirt cheap, impedance balanced outputs when connected to a true balanced input, will provide the adequately balanced line to accommodate good CM noise cancellation.

Note: If receiver is not true balanced but merely differential input, or even just single ended, adding an impedance balance resistor in series with the output low side delivers little benefit and could even slightly degrade signal integrity depending on the specific receiver topology.

It's not surprising that someone promoting balanced inputs would encourage output designers to provide him with a signal that will allow his inputs to work properly. Using dual active output drivers leaves less money in the budget to buy his input chip sets.

IMO the odds are better that impedance balance resistors will help or be neutral to signal integrity, so go for it. It won't break the bank. If you need to impress customers the active dual outputs gives you a nice hot peak output signal to tout on your data sheets.

JR

Well i tried to record the differencies it made using capacitors but it wont work near a computer or my 002R...to much RF interferencies...

Gonna try to build it with 48 phantom circuitry.

I had a look at THAT's website, and the App note #109 looks for all the world like the Valley Transamp circuit with the addition of a few refinements like the DC offset nulling servo. This looks like it would make an excellent preamp. If someone gets a chance to try this circuit, let me know how it performs.

Just out of curiosity, have you considered using the THAT1646 in conjunction w/ your 1500 series pre?

1646 is a nice balancer..i don't have any yet but i'm gonna order some!

Wait a bit, I need to get some THAT chips too.

jim