Here's a handy calculator that will illustrate DF96's point quite nicely:
KEMET Software
This really is pretty simple stuff.
KEMET Software
This really is pretty simple stuff.
In the issue here the self resonance due to lead inductance is not as clear. The capacitor has one side connected to a shield cable of unknown length. The other end may be modeled as a ground since it is going hopefully to a chassis.
The result is that the self resonance frequency can be much lower than in a more typical application such as mounted on a circuit board.
It shows otherwise. Even adding in 3nF for the leads, the impedance at 150MHz is a whopping 2.5 ohms.
Nordholt & van Vierzen MC stage?
Anyone ever tried their MC stage? I have uncovered something very special in my junk box, possibly the only two diffpairs left in existance of the transistors mentioned in the paper Low Base Resistance Integrated Circuit Transistor, Willemsen and Bel, 1979. They were a gift from the authors to Barrie Gilbert and he gave them to me (>30yr ago). There are also two op-amps in a fixed gain of 50 that I was told were made for Ortofon, but I have never seen any evidence of that.
The transistors have rbb' of 1.5 Ohms, so paralleling a pair could ge me down to sub .2nV (sort of a bipolar version of the crazy Interfet JFET). They are in nice old fashion military gold/ceramic packages so 50-100mA Ic is no problem. The authors claim "significant" improvement in the Nordholt & van Vierzen circuit, but offer no details.
Anyone ever tried their MC stage? I have uncovered something very special in my junk box, possibly the only two diffpairs left in existance of the transistors mentioned in the paper Low Base Resistance Integrated Circuit Transistor, Willemsen and Bel, 1979. They were a gift from the authors to Barrie Gilbert and he gave them to me (>30yr ago). There are also two op-amps in a fixed gain of 50 that I was told were made for Ortofon, but I have never seen any evidence of that.
The transistors have rbb' of 1.5 Ohms, so paralleling a pair could ge me down to sub .2nV (sort of a bipolar version of the crazy Interfet JFET). They are in nice old fashion military gold/ceramic packages so 50-100mA Ic is no problem. The authors claim "significant" improvement in the Nordholt & van Vierzen circuit, but offer no details.
Last edited:
Anecdote. Back when designing multimedia/computer-oriented powered speakers, the assignment was to provide a synthetic soundfield out of two satellite speakers from a multichannel output soundcard. The particular ultra-simple scheme of Norris was selected (a patent that Harman abandoned, despite it already getting office action --- but Gina had deposed Druckrey, and was appalled at the money spent on patent lawyers, so a whole host of patents languished, including some decent utility ones).
I suggested that it would be nice if, contrary to Norris' scheme, which had a means of transitioning from synthetic surround to mono, we provided a way to go from synthetic surround to stereo. I then had to figure out how to do it.
It turned out that a version of the "phase splitter", in this case using complementary feedback pairs of bipolars, was part of a felicitous solution. Early in the signal chain the two polarites were produced and thereafter processed. The board carrying the processor and potentiometers for adjustment, accessible to the customer on the top of the "subwoofer" was a simple double-sided one with ground planes put down wherever possible. But the customer (Dell) did not place any particularly stringent conditions on RF immunity, so we just hoped for the best. As usual the schedule was absurdly tight.
At a trade show the system was set up in a hotel room. At one point one of the sales guys was using his cellphone in proximity to the speaker. I was stunned --- there was no trace of audible demodulation, which otherwise was always present in earlier powered speakers.
And yet bipolars in the front end? And a plastic case, with no shielding whatsoever? At least in terms of that path for pickup, the likely reason was that, for purposes of achieving low noise, the transistors were being run at a rather high bias current and consequently had appreciable gain-bandwidth, and as well high base currents, particularly as compared to, say, opamps commonly used for audio. So the rectification/envelope detection propensities of the transistors were mitigated. There may have been some luck with the layout too.
The particular variable surround synthesis scheme and variations on it are patented, although frankly I never cared much for how it sounded 🙂
Brad
I suggested that it would be nice if, contrary to Norris' scheme, which had a means of transitioning from synthetic surround to mono, we provided a way to go from synthetic surround to stereo. I then had to figure out how to do it.
It turned out that a version of the "phase splitter", in this case using complementary feedback pairs of bipolars, was part of a felicitous solution. Early in the signal chain the two polarites were produced and thereafter processed. The board carrying the processor and potentiometers for adjustment, accessible to the customer on the top of the "subwoofer" was a simple double-sided one with ground planes put down wherever possible. But the customer (Dell) did not place any particularly stringent conditions on RF immunity, so we just hoped for the best. As usual the schedule was absurdly tight.
At a trade show the system was set up in a hotel room. At one point one of the sales guys was using his cellphone in proximity to the speaker. I was stunned --- there was no trace of audible demodulation, which otherwise was always present in earlier powered speakers.
And yet bipolars in the front end? And a plastic case, with no shielding whatsoever? At least in terms of that path for pickup, the likely reason was that, for purposes of achieving low noise, the transistors were being run at a rather high bias current and consequently had appreciable gain-bandwidth, and as well high base currents, particularly as compared to, say, opamps commonly used for audio. So the rectification/envelope detection propensities of the transistors were mitigated. There may have been some luck with the layout too.
The particular variable surround synthesis scheme and variations on it are patented, although frankly I never cared much for how it sounded 🙂
Brad
Yow 1.5! That's low. As I may have mentioned I exhorted a friend to buy some Toshiba 2 ohm parts, and along the way got 200-piece bags of each polarity from him for myself. They will come in handy someday, if I live long enough (my nearly 1k pieces of 2SK389V parts are also awaiting use, somewhere).
Do you have links or other specifics on Willemsen and Bel?
Brad
P5
There was an Elector article using BF*** parts, I guess this was best for 1979 and integrable but not hard to do with only 4 or so paralleled modern devices. The link got tiny, P5 above. Not a very special circuit, giant electrolytic at input yuk.
Last edited:
If you want a serious RF bypass solution, contact Scott Wurcer. He helped me 28 years ago with his suggestions, based on REAL measurements.
Thanks! Yes, via a search I also see that W & B are available but cost money, and I really don't need to know how they achieved the 1.5 ohms as I am not about to make transistors.
Yes, the Nordholt & van Vierzen schematic is not, errrrr, very appealing.
Brad
Where does that number come from? Your reference from a few posts back is down the memory hole?
edit: I calculate maybe 20nH worst-case. That brings up the series impedance to a mind-boggling 12 ohms at 100MHz. Ed's erstwhile reference shows a smaller inductance. Who cares? 12 ohms is far and away sufficient.
The reference was for a capacitor in a typical circuit environment. I showed it to demonstrate that small leaded ceramic caps run out of bypass usefulness at fairly low frequencies. That showed a 1 nF cap the initial question was for a 10 nF. So we have more capacitance and much more lead length.
One end of the capacitor is connected to the case via some small (<5 nH) lead. The other end is connected to the shield. Pick your best guess for a 72" lead length (or 36" if you prefer.) That is where the resonant circuit would be.
If we were to treat the shield as just a freestanding length of wire .09" dia by 36" L the inductance would be 1 uH (CRC formula). Although the shield is hollow the inner conductor is not connected via a low impedance path to ground.
As to the loss for 12 ohms 20log(110/12) = 19.244 db. Which would help but really is not a guarantee of EMI proofing a circuit.
Small ceramic capacitors are effective for AM commercial radio filters. AM is the worst problem for audio gear. But you will not pass current EMI standards if that is the primary tool in your kit.
You're all over the map here. We're talking about a ceramic cap from input connector to chassis. 1/4" leads, max. If you put 72" leads on it, you're designing for Martians.
Sy,
We probably can agree if you have an RCA connector mounted in an insulated bushing with nothing connected to it, RFI will not be a problem even with a 10 nf ceramic capacitor with 1/4" leads.
Now what is the effective lead length when you plug something into the jack?
We probably can agree if you have an RCA connector mounted in an insulated bushing with nothing connected to it, RFI will not be a problem even with a 10 nf ceramic capacitor with 1/4" leads.
Now what is the effective lead length when you plug something into the jack?
Still the same. That's the whole point of the cap to the chassis; that's why they call it a "bypass." I hope you're not suggesting to attach one lead to the sending end, the other to the receiver chassis.
That was your worry, not mine; I was just putting some numbers to it to try to figure out what you're on about. Bypassing the input plug to the chassis with a small ceramic cap with reasonably short leads is a standard method, and it's standard because it's effective. I think my Collins S-Line even used the same method. I put some numbers to your example, and that led to less physically reasonable examples.
- Status
- Not open for further replies.