John Curl's Blowtorch preamplifier

[Bob Cordell]

And not all pre-amps have a Zo of 50 Ohms.

Cheers,
Edmond.

Hi Edmond,

This is true, but a transmission line needs to be properly terminated at only one end to behave.

Bob

 

[PHEONIX]

Cordell power amp Jfet/BJT input

Hello Bob

It would be interesting if you could change the input to BJT and confirm which of the two inputs gave lower THD performance.

The reason I make this inquirey from what I have gathered from Scott Wurcer spice models of JFets are "usually" not as good as BJT spice models.

Regards
Arthur

 

[syn08]

This is true, but a transmission line needs to be properly terminated at only one end to behave.

Are you sure Bob?

 

[hitsware]

>This is true, but a transmission line
>needs to be properly terminated at
>only one end to behave.

I think that depends on whether it is
being run at 1/4 wave or 1/2 wave or
?/wave ..... and what you call
"properly terminated" and "to behave" ......

 

[AndrewT]

Re: Re: Re: Re: Re: Re: RFI

........... bandwidth of the amplifier limited by ..........passives than by the active closed-loop frequency response of the amplifier-proper............... In any case, I believe it is usually best that the frequency response droop at 20 kHz be kept to about 0.1 dB or less.

One thing that I like to do is right at the RCA connector where the signal comes in, run it through a series 49.9 ohm resistor and then a shunt 100 pF capacitor to the connector ground so that the interconnect is effectively terminated in 50 ohms at very high frequencies. This, of course, is a compromise, as not all interconnects have a characteristic impedance of 50 ohms,

hear, hear.
or even a 100r feeding 100pF @ the input terminal.

I also advocate that the LF end be filtered similarly, i.e. aim for F(-0.2db)<=20Hz

 

[Edmond Stuart]

Hi Edmond,

This is true, but a transmission line needs to be properly terminated at only one end to behave.

Bob

Hi Bob,

So the other end needs to be improperly terminated?
And how, with 0 Ohm, 100 Ohms, 1kOhm 10kOhm?
Please explain.

Cheers,
Edmond.

 

[PMA]

You joking?

It is enough to terminate e.g. RG-58 by series resistor 50 ohm at the input, cable output left open

OR

to feed it from voltage source at the input and output terminated by 50 ohm.

Then you get only small difference from ideal response, caused by real cable properties. Try it. You can also simulate - MC has transmission line macros.

 

[Edmond Stuart]

Pavel, I'm not joking, just asking a question. According to post#7159, Bob proposes to terminate the cable only at the side of the power amp with 49.9 Ohm. So, I'm asking what to do with the other side.

BTW, "to feed it from a voltage source at the input" is impossible. Every signal source has an output impedance. Moreover, I think it's not a good idea to connect a coax cable directly (without a series resistor) to an output and loading it with a (unknown) capacitance.

 

[PMA]

Bob is right, regarding HF termination (series 49.9 at the cable input). The end left open, i.e. 10k input impedance of following component is like open.

This is true only for 50 ohm coax. cable, RG-59 would need 75R etc.

For this reason, I came with

http://web.telecom.cz/macura/buffer_en.html

years ago.

 

[AndrewT]

Bob is right, regarding HF termination (series 49.9 at the cable input). The end left open, i.e. 10k input impedance of following component is like open.

is right at the RCA connector where the signal comes in, run it through a series 49.9 ohm resistor and then a shunt 100 pF capacitor to the connector ground so that the interconnect is effectively terminated in 50 ohms at very high frequencies.

but Bob is proposing that we terminate the receive end with 50r.
Bob makes no mention of the source end of the interconnect.

 

[Edmond Stuart]

Pavel,

This is what Bob was saying, while he was discussing input filters:

"One thing that I like to do is right at the RCA connector where the signal comes in, run it through a series 49.9 ohm resistor and then a shunt 100 pF capacitor to the connector ground so that the interconnect is effectively terminated in 50 ohms at very high frequencies. "

So, what do you think he's talking about? The pre-amp side or the power amp side? (I think the latter)

 

[Edmond Stuart]

Thanks Andrew. At least one guy who can read.

 

[PMA]

Oh - I see. He proposes only to terminate for HF (50R + 100pF in series). Then you have to feed the RG-58 from zero ohm or 50 ohm impedance.

 

[ingrast]

.....

So the other end needs to be improperly terminated? ....

2 issues here,

One wants to fool the cable to think it is infinite so any signal travelling along will not find discontinuities and be partially reflected. Loading the far end with the characteristic impedance does the trick.

For the source end, 2 further issues. By the same token one wants to avoid any reflected - or backward signal fo any reason - to be further reflected which yields the familiar ringing in high speed unterminated circuits. Providing the characteristic impedance at the source does the same as before.
Morover, it can be shown - though it is not an issue for low power audio - that maximum power transfer is achieved when source and load impedance are the same, this is a further reason in RF circuits to care match source and transmission line impedances.

Rodolfo

 

[Edmond Stuart]

Pavel, I'm glad you finally see the light.

So 50 ohm impedance also at the other side and, as I said before, not zero. (see your own R10=50 Ohm, according to the link you've just posted)

 

[PMA]

Just for HF, zero impedance at cable input and 50R + 100pF as cable termination would work as well.

 

[Edmond Stuart]

No.

 

[lumanauw]

This article agrees with PMA's method :
http://www.6moons.com/audioreviews/dartzeel2/preamp_3.html
The graphs are interesting 😀

 

[Charles Hansen]

Once again, off topic.

But since we are there, I am surprised that there is so much misunderstanding about cables and terminations.

It is only necessary to terminate one end of a cable with the characteristic impedance. Bob Cordell's method is fine (terminating at high frequencies only), as the cable doesn't even have a constant (characteristic) impedance at low frequencies.

You can terminate the source or the load or both ends. Slightly different things happen in each case. But in all cases you will avoid the problem of multiple reflections that create a signal that looks like "ringing".

The best book to read on this topic is "High Speed Digital Design" by Howard Johnson:

http://www.amazon.com/High-Speed-Digital-Design-Semiconductor/dp/0133957241/

 

[Charles Hansen]

This article agrees with PMA's method :
http://www.6moons.com/audioreviews/dartzeel2/preamp_3.html
The graphs are interesting 😀

As usual, there is some misinformation here. (No surprise -- after all, this is the internet we are talking about.)

The formula given for characteristic impedance is incomplete. The full formula includes the resistance of the wire and the conductance of the insulator. At high frequencies (above the audio band), the inductance and capacitance dominate and it is acceptable to use the simplified formula. Usually.

But in the case of the OCOS cable, this is not OK. The OCOS cable deliberately introduces a controlled conductance into the insulator to create a characteristic impedance of 8 ohms. This is to match the loudspeaker, and also this technique extends the effective range of the characteristic impedance down into the audio band. Ignore the incorrect figure of 60 ohms given in the article.