Does a single ended signal see the common mode input impedance of 10M?
I'm trying PLLXO with Mod86, XO~100-300Hz. Need Zin to figure out the part values.
Thanks,
Rich
I'm trying PLLXO with Mod86, XO~100-300Hz. Need Zin to figure out the part values.
Thanks,
Rich
In a single-ended or pseudo-differential configuration, the input impedance should be the differential input impedance, so 48 kΩ.
If your filter cutoff varies greatly with the load impedance, you can always add an LME49710 buffer. That'll give you Zload of GΩ for your filter and Zout = 0 Ω (approx) for driving the THAT1200.
Tom
If your filter cutoff varies greatly with the load impedance, you can always add an LME49710 buffer. That'll give you Zload of GΩ for your filter and Zout = 0 Ω (approx) for driving the THAT1200.
Tom
Thanks for the quick reply Tom.
My R1 is 3650ohms, as low as I can go with 37ohm opamp source. With 10MOhm load, my R2 is 36.5kOhms. With 48kOhm load my R2 is 157kOhms, too high insertion loss.
BUT... it would work nicely between THAT1200 and LM49710. And then I could use the balanced input too.
That perfectly buffered real estate is very tempting to invade with VC, filters, etc. But must the link between 1200 and 49710 remain short and quiet for stability?
Thanks
Rich
Last edited:
There's no stability concerns with tapping into the MOD86 between the THAT1200 and the LME49710. I'd lift pin 6 of the THAT and connect the filter there. Just be careful with the grounding. That's probably the biggest concern as the connection at that point is single-ended and any ground error voltage becomes input signal to the composite amp.
Tom
Tom
I'd use the closer one of those three.
R13 refers to Rev. 1.0 of the board. On Rev. 2.0, that resistor would be R14.
Tom
R13 refers to Rev. 1.0 of the board. On Rev. 2.0, that resistor would be R14.
Tom
It's done, actually: THAT Driver: Differential line driver / preamp
I should probably start a vendor thread for it. Sorry. I've been a bit busy moving (Seattle -> Calgary) so my focus has been elsewhere. I'm 99.99 % back to normal though. I'm ready for orders and am starting to have enough flex in the schedule to allow for new product development also.
Tom
Nice!
From the graphs, it seems to be very transparent. Can a 25K pot or R2R ladder be installed on the input?
Thanks
Do
Absolutely. Personally, I'd probably go for a 10 kΩ pot (lower resistance -> lower noise), but 25 kΩ will work as well.
Tom
I'd have some questions (about the input filter section) that could go in it: 😀
I see a dual ferrite bead arrangement per channel. Did you compare this to common-mode chokes (e.g. Würth WE-SL2, ignoring SMT vs. TTH for a moment)?
You use discrete diode arrangements. Did you compare this to integrated solutions?
I understand that you go by the application note recommendations by THAT, but have to ask out of curiosity. 😉
Cheers,
Sebastian.
The chokes are actually the RFI filter on the output. They prevent RF from being injected into the circuit via the XLR cable. I didn't compare with common-mode chokes as I'd like the filter to work both on differential mode and common mode signals.
Not sure what you mean by "integrated solutions". Multiple diodes in the same package? I suppose I could have used a bridge, but I don't see how that would be an improvement over the 4x1N4007s I have in the circuit already. A diode bridge takes up less area, but is typically wider than two diodes side-by-side, so the bridge doesn't necessarily save board space.
Tom
Hey Tom, am I wrong in thinking that those chokes are typically horribly non-linear and would adversely affect the associated output signals?
Edit: I'm guessing it's a moot point though as there isn't much distortion visible in your measurements 😛
The chokes/inductors behave differently at different frequencies.
They are very low value resistors at DC.
If perfect inductors they behave as infinite impedance at light frequencies.
In between they are a combination of (imperfect) inductor plus series resistor and a tiny bit of parasitic capacitance.
The trick/skill of the Designer is to select a choke that behaves as a resistor where you want the signal to pass and yet behaves as an inductor where the frequency is above the wanted signal.
A common mode bead around the signal Flow and signal Return lines behaves as a low value resistor for all the signal passing to and fro along the line/resistor
They are very low value resistors at DC.
If perfect inductors they behave as infinite impedance at light frequencies.
In between they are a combination of (imperfect) inductor plus series resistor and a tiny bit of parasitic capacitance.
The trick/skill of the Designer is to select a choke that behaves as a resistor where you want the signal to pass and yet behaves as an inductor where the frequency is above the wanted signal.
A common mode bead around the signal Flow and signal Return lines behaves as a low value resistor for all the signal passing to and fro along the line/resistor
I'm a bit confused by sek's question as well. It might be getting at single package EMI and filter networks (RC, LC, CLC, LCL, differential, common mode, pick your combination of flavours), some of which include ESD diodes. Those are mostly fine pitch surface mount, corner at undesirably high frequencies for audio, and I'm not aware of any with larger diodes intended for line clamping or such---usually the application is USB or such, though ON and NXP offer a for mobile headsets. The most suited to this application I know of is ON's IP4048CX5.
Didn't recognize they are sitting at the output, but since it's a line level application the requirements are essentially the same for both in- and output filtering/protection.
What difference and/or what harm would you expect a common mode choke to do in single ended mode (i.e. other leg floating).
What are the discrete chokes doing to the balance (with real world interconnects/EMI)?
ESD protection packages come in a multitude of shapes and configurations, a diode bridge just being one of them.
They essentially are convenient solutions to fit many diodes of various types onto the board without adding too much capacitance and inductance (i.e. response lag/spikes). Worth looking into...
I wasn't referring to filters at that point, hence I don't understand the stab about corner frequencies. But I agree that a USB filter isn't ideal for audio.
It's correct that most integrated solutions come in SMD.
But – as a general thought – since virtually all new integrated semiconductor products are packaged as SMD I somewhat tend to associate disapproving SMT with reluctance towards progress. Many recent semiconductors objectively offer improved performance over their predecessors, yet aren't available in DIP/TO packages any more. 😉
Thanks,
Sebastian.
Not sure it's 75% but yes, DIYer demand runs to through hole builds and assembled surface mount modules. Personally I find assembling surface mount faster and easier than through hole---yeah, my builds are medium pitch surface mount (e.g. 0805s and SOICs) or finer---but my experience of several maker communities is there's a definite fear/uncertainty/doubt speedbump over trying surface mount assembly which doesn't exist for leaded parts. Some of this is just that one doesn't usually breadboard with surface mount parts, some of it is lack of tooling such as temperature controlled soldering irons or reflow ovens, and some of it just seems to be the need for a bit different mindset in regards to keeping track of smaller parts.
Sebastian, the value of a lower corner frequency is rejection of noise starting at a few hundred kHz and a well established stopband at tens to hundreds of MHz where integrated EMI networks corner. What integrated networks offer is deeper GHz stopbands due from lower parasitic inductance. Handy for keeping noise from CPUs and other fast digital logic on the same board from radiating off cables plugged into mobiles, tablets, ultrabooks, and whatnot. Not much of a scenario for the THAT Driver or Modulus, though.
There are a number of surface mount diode arrays I like for such lighter duty purposes when the line driver or receiver doesn't integrate them. Through hole is nearly all TO-220 other large packages with few smaller options rarely being cost effective; at that point distinction from axial parts starts to look a bit blurry to me.
Sebastian, the value of a lower corner frequency is rejection of noise starting at a few hundred kHz and a well established stopband at tens to hundreds of MHz where integrated EMI networks corner. What integrated networks offer is deeper GHz stopbands due from lower parasitic inductance. Handy for keeping noise from CPUs and other fast digital logic on the same board from radiating off cables plugged into mobiles, tablets, ultrabooks, and whatnot. Not much of a scenario for the THAT Driver or Modulus, though.
Keep in mind the external 16x6 protection network THAT includes in the datasheet and (near as I can tell) replicated in the THAT Driver is intended to crowbar a phantom microphone supply, meaning the protection diodes need to discharge bypass capacitance at 48V. 1N400xs are a good choice for this purpose in a through hole build like the THAT Driver. Refer to the datasheet and you'll see THAT indicates the on die protection networks "suffice for many situations in the field" which, considering the robustness requirements of pro audio, is their way of saying they've addressed the problem you're concerned with.
There are a number of surface mount diode arrays I like for such lighter duty purposes when the line driver or receiver doesn't integrate them. Through hole is nearly all TO-220 other large packages with few smaller options rarely being cost effective; at that point distinction from axial parts starts to look a bit blurry to me.