Looks like a very cheap way to do it. I'd not split the gates in the logic chips the way its shown there though - keep the linear operating one (with the fb resistor) in a different package from all the output driving ones.
That's probably a good idea, it might not make much difference, but without measuring you can't tell, and the chips are cheap.
w
Cheap as chips we say here, that means French fries. Cheap as French fries doesn't have that je ne sais quois though.
w
Cheap as chips we say here, that means French fries. Cheap as French fries doesn't have that je ne sais quois though.
Thanks for posting it.Since I'm not up to speed with logic why might this circuit be preferable to using the AD811?
Here is the data sheet. Go to page 14 where they show how to split to two outputs and note they claim 40dB isolation at 5MHz when splitting to two outputs.
http://www.analog.com/static/imported-files/data_sheets/AD811.pdf
The isolation would be better if the amp approached 0 ohm output but it's pretty dang good as it is. You may have noticed that I said the 3 output DA in the Nova telecine used a separate 811 for each output to get the best possible isolation.
There is nothing wrong with your multiple resistor solution apart from isolation between cables and a serious signal loss. Do you consider that acceptable? The good thing about a DA is virtually no interaction between the outputs and no signal level changes.
At the time the Nova was being designed it was done by a company called Digital Audio and Video (DAV) and the machine was called the Cineglyph. My initials (G²) are on at least 17 of the PCBs in that machine as Dave Walker allowed me to sign my work on the component layer in copper. The signal system is completely DC coupled except for 1 in line sample and hold capacitor on each of the RGB channels.
G²
The HC04 will work. The impedances are aren't right - they're close. The levels aren't right - maybe not even close but likely will work. Preferable? No. Works? Yes but you're relying on the DAC designer to have done an excellent job to accept sloppy signals without complaint.
You should really hang a scope on the inputs and outputs to see what really goes on in a transmission line. Just the terminator at other than the physical end point can cause unwanted reflections and ringing. The 811 solution will give an extremely accurate replica of the original signal complete with any distortions that may be present. Isn't that what you want? I would think that is the best way to compare DAC sonic performance and not how well the DACs fix poor input signals.
G²
I just noticed something. You say you'll scale it back to unity gain. When dealing with DAs and 75 ohm inputs and outputs, 'unity' is 2, not 1. Remember, you need the 75 ohm resistor to load the transmission line correctly on the input. To drive the line you series feed a 75 ohm resistor and when you terminate the destination end, you've just formed a 2:1 divider with the series 75 ohm resistor and the termination to ground. To maintain 'unity', the gain must be 2.
G²
To terminate use two 150 ohm resistors for a y cable. They are seen by the signal in parralel, thus =75 ohm.
I see Stratus has not convinced you that he has done nothing more than show how to amplify the signal.
Your 150r suggestion is not correct.
Each output leg and each input leg must be terminated at both ends with the same impedance as the characteristic impedance of the cable.
For 75ohm Coax the far end must have a 75r grounded load.
The near end must have a 75r source impedance. If the source has a 0r0 output impedance and there is only one cable then a 75r is inserted between the coax core and the zero output impedance feed point.
When you have more than one cable attached, every coax core must be fed from an effective source resistance of 75r. If one of those sources is 5r or 0.1r or 21r7, then that combined with all the other coax connection must be taken into account when determining the source resistance that needs to be added to the coax core you are considering.
Your 150r suggestion is not correct.
Each output leg and each input leg must be terminated at both ends with the same impedance as the characteristic impedance of the cable.
For 75ohm Coax the far end must have a 75r grounded load.
The near end must have a 75r source impedance. If the source has a 0r0 output impedance and there is only one cable then a 75r is inserted between the coax core and the zero output impedance feed point.
When you have more than one cable attached, every coax core must be fed from an effective source resistance of 75r. If one of those sources is 5r or 0.1r or 21r7, then that combined with all the other coax connection must be taken into account when determining the source resistance that needs to be added to the coax core you are considering.
Unfortunately, no. You're thinking shielded cable, not transmission line. They look similar and in fact they are but transmission line is made to a specific impedance. 75 is the most common but there is 93 ohm, 50 , 51 and others. EACH length of cable must be terminated at both ends with the correct impedance to behave properly. You can measure the characteristic impedance of any shielded cable and use it that way but the values are likely to be very peculiar and not very good performance.
The size of the center conductor vs shield diameter and the dielectric properties of the insulation determine the impedance. The centering must be maintained as well which is why transmission line coax costs more.
G²
74HC04 - $0.50
AD811 - $7.00
The 811 is overkill, the concerns about matching are exaggerated. Few cables are toleranced better than +/- 1.5 ohms, the terminations of most DACs are RCA connectors which have an impedance of ~30 ohms, followed by a bit of PCB track, followed by a 75R resistor, but the rise times of the signals are such that these terminations do not result in significant reflections, otherwise BNC or other superior connectors would be required, as they are with faster circuits. RCA are standard with SPDIF, XLR with a AES/EBU.
The output impedance of the HC circuit is 72R, you can add a 3R3 resistor in series with each output if you care. As Rod Elliott says 'The error is too small to worry about'
There is certainly no doubt that with the 74HC the outputs are isolated.
The AD811 is a good solution, but better than the 74HC? The difference just isn't worth arguing about, and even if you use 2 x 74HC, the AD811 is never $6 better.
The one thing nobody noticed is that the circuit as I have drawn it is inverting, so you might want to take out U1B, or insert another gate after it would be better if you use 2 x 74HC.
w
"your" circuit - it's not your's in any way - is exactly the configuration, which is producing this kind of reflection on it's input, shown here below.
It is not +- 1.5 ohm difference. Yes, the reflected wave really is 80 % of the incident wave on the screen there. For > 4nsecs.
Your - not, Eliot's receiver will produce less of it, ~like 30% of the incident wave.
Plus never really reaching 75ohm, like in the shot - so it is everything, but 75ohm.
Beautiful, refined design really. Not worth the price to listen to people like stratus46, when there are THIS capable RF experts at hand.
Ciao, George
Ps.: Vertical is 500mrho, horizontal is 5nsec in this TDR plot
It is not +- 1.5 ohm difference. Yes, the reflected wave really is 80 % of the incident wave on the screen there. For > 4nsecs.
Your - not, Eliot's receiver will produce less of it, ~like 30% of the incident wave.
Plus never really reaching 75ohm, like in the shot - so it is everything, but 75ohm.
Beautiful, refined design really. Not worth the price to listen to people like stratus46, when there are THIS capable RF experts at hand.
Ciao, George
Ps.: Vertical is 500mrho, horizontal is 5nsec in this TDR plot
Attachments
Last edited:
No, it's not mine, I credited Rod Elliott with it. So what's your point?
What isn't?
So why are you showing us this? It's not the receiver in the circuit. So what is your point?
What?
No-one can make sense of what you are trying to say.
Write it out again JosephK, this time try to keep your temper. The trouble with most of what you write is that you are relying on extrapolations of what you have been able to measure. This is part of the problem. You get access to a nice oscilloscope, so you measure what you can measure with an oscilloscope.
Produce a couple of eye diagrams of recovered clocks showing significant jitter or an FFT showing a difference or even tell me you got somebody to change the setup without you looking (as in the case of the Hiface modification, for example) and you could hear the difference, and I will be much more interested in what you say.
But no, all you have to offer is 'I am sure I can tell the difference' 'here is a picture of one that is like it'.
You have produced a plot of some device showing a reflection, but not the device in question. Why?
You repeatedly attack me on a personal basis and attempt to discredit me because I have disagreed with you. Please try to refrain from this in future. What is at issue here is not my competence as an RF engineer, but the relative merits of the circuits.
w
Oh, looking at your plot, you say 5nS/div, so the rise time is 1nS or less.
Last edited:
I had been only talking of the special merit of this special circuit shown by You,
and saying that it is far from a good termination, on it's input.
If that, a good termination and reflections created are important or not, was not my subject this time. Though, in this thread quite some other persons were discussing exactly that.
and saying that it is far from a good termination, on it's input.
If that, a good termination and reflections created are important or not, was not my subject this time. Though, in this thread quite some other persons were discussing exactly that.
Last edited:
Would like to add, that while the suggestion of Wakibaki has issues, I can only agree with everything what Stratus46 had to say up to now.
The star of resistors suggested by AndrewT, is also a working solution.
It can be done to a level where only the crosstalk / attenuation issues remain.
Attenuation is 6db. Though also the crosstalk is 6db.. Pro: it's incredibly simple and do not introduce power supply modulation effects..
Ciao, George
The star of resistors suggested by AndrewT, is also a working solution.
It can be done to a level where only the crosstalk / attenuation issues remain.
Attenuation is 6db. Though also the crosstalk is 6db.. Pro: it's incredibly simple and do not introduce power supply modulation effects..
Ciao, George
- Status
- Not open for further replies.