John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
JN

PM me a shipping address. I will send you a few coils to measure so we can compare notes. I think you are drifting off the issue of disagreement.

We both understand there is more than one issue affecting signals in a cable. The issue is inductance and if it changes in a coiled cable.

So inductance bridges at high noon, stranger!

ES
 
I have a question to those who might know the details better - John, Scott, Brad..

What's the difference, if there is any, in Jfet manufacturing process that makes the transistors from the same process (say 51, National semi) switching ones in one case (J111) and low noise in other case (PF5101)? Quality control? Some ommited or extra process? Different wafers? Cleaninness? Different manufacturing line?

Thanks!
 
Channel separation measurement for JN

I have made one more channel separation measurement, especially for JN. Now, the whole chain of Preamplifier + PA4 power amplifier was measured. With cables and everything, of course. Then, a measurement of Preamplifier alone was done. One may see that the power amplifier has not changed the measurement result! Finally, as a reminder, channel separation of power amplifier PA4 alone is shown. The conclusion is that it is a preamp that defines channel separation, in case that power amplifier has a proper dual-mono layout and proper wiring.
 

Attachments

  • channel separation Preamp + PA4.PNG
    channel separation Preamp + PA4.PNG
    22.2 KB · Views: 272
  • channel separation Preamp only.PNG
    channel separation Preamp only.PNG
    21 KB · Views: 268
  • channel_separation PA4.PNG
    channel_separation PA4.PNG
    21.8 KB · Views: 268
I have a question to those who might know the details better - John, Scott, Brad..

What's the difference, if there is any, in Jfet manufacturing process that makes the transistors from the same process (say 51, National semi) switching ones in one case (J111) and low noise in other case (PF5101)? Quality control? Some ommited or extra process? Different wafers? Cleaninness? Different manufacturing line?

Thanks!
You need more information, the noise can depend on channel length within the same process. Other things, gettering could be left out, starting matierial maybe, different fab maybr too, but I don't think anyone can afford diirty fabs today.
 
You need more information, the noise can depend on channel length within the same process. Other things, gettering could be left out, starting matierial maybe, different fab maybr too, but I don't think anyone can afford diirty fabs today.

I can't find PF5101 DS so can't tell if it is actually same die. Seriously cold here, SY must be experiencing real winter for once. ;)
 
Elektroj, it is just selection. Some die come out quieter than others. For SPECIAL low noise selection, this is an extra step over just being a switch.
Toshiba had been making better devices for more than 30 years, now, so we just use the 2SK170 for most low noise inputs, getting a little input capacitance for quieter operation and more Gm.
 
J.N.

My recount of the events is that there was a statement that coiling a microphone coil changed the sound and this was explained by a change (increase) in inductance.

You stated there is no change in inductance on a twisted pair cable.

I ran measurements that showed a small difference.

You complained that the comparison was invalid because I was not accurately measuring the inductance.

I repeated the test showing the actual inductance and the change. It was very small as previously mentioned and as expected. It was on the order of .5% for a twisted pair and 5% for twisted pair with the shield shorted to one side. (This data was in the image captions. If you hold your mouse over the image the caption appears.)

You now have done your own test, but the graphs you have presented were not of sufficient resolution to show the differences.

The very interesting result was the change in your measurement of resistance. I would expect skin effect to come in around 40-50 kHz for 22 gauge wire. (The wire diameter not really being exact.) Do you have an explanation for this or just an educated guess.

Now the result of wire changing characteristics due to coiling or handling are not exactly unknown. Cat 5 cable (It is a cable not a wire!) has twisted pairs. The improved version adds glue to keep the twists the same. That is because as you coil or bend the wire it stress the twists and adds a bias to the pairs depending on their location in the cable.

I also measured crosstalk between unconnected cables. You seem to believe this is not due to inductive coupling. What do you think causes this effect? (True it is very low, as low as -100 db/100' in some cables.)

Another issue involved is overall cable diameter. Portable microphone cable is often wound with two cord fillers to make the cable rounder. This does tend to increase the distance between adjacent coil wraps.

Of course some cable uses what is called "Star-Quad" where there are four wires twisted and these are then wired with two conductors becoming each side of the twisted pair. This construction provides better noise rejection. The shield stays the same as in other constructions.

Now why a 1000' foot coil of cable would produce poor results with a measurement microphone is interesting. The "Telegraphers Equation" would expect that a 120 ohm source and a 120 ohm load (The cable used has a nominal impedance of 110-120 ohms) should show a flat frequency response but with some uniform attenuation.

The first pass answer would be the microphone source was not 120 ohms but actually lower and the termination impedance was higher. A second issue would be slew rate limiting. At 50 pf/foot there would be 5 nF of capacitance, as the measuring microphone was phantom powered it most likely had a very limited current output. During testing high level signals are used to get adequate S/N ratios. So that seems to be a likely culprit. Now you used a stainless cylinder to test for solenoid effect. If I can find a metal spool I will try that to see what effect there is. Virtually all the wire comes on plastic spools today, so that may be a bit more difficult.

ES
Since I started this coiled mic tangent . Some important field data does come to mind. Cincinnati music hall where the event happened is in a strong FM field about 1.5 to 1.8 volts rf in the house at 90.9 mhz . the coil was at least 250 feet if I remember correctly . Though I will not disagree with the theory that a coiled mic cable should not have added inductance what if the helical coil as an antenna for strong fm signal. When we take thing out of context the wrong conclusion . In this case the coil acting like an antenna in a strong RF field may have been injecting that RF in the system . When the antenna was detuned the interference was reduced . So a small increase in inductance in and of itself being measured small could have caused a problem not proximal circuit designed. Look forward to you view on that limited possible cause.
 
I just finished a phone call with Dick Sequerra, a man wizened by experience, education and realizing audio products that 'work' well, above and beyond just 'specs'.
We agreed that to be extra successful, an elegant, but simply executed design is best. Adding a lot of extra parts to compensate for problems that could have been eliminated by using better parts and materials in the first place is less than optimum. Now, I know that many here get excited by complex and 'sophisticated' circuitry, that usually contains lots of inexpensive parts, configured to get good static numbers, but I find the simple and elegant design approach to sound better.
 
Coiling a cable is more likely to prevent it acting as a VHF antenna, although this could depend on exactly where the coil is. The coiled part acts as a common-mode choke.
True what encountered was a very high to signal ratio field 1.5 volts RF at 90.9 mhz with a signal well below that . My view was that a tuning effect may have happen . I am guessing at that . It that general area phono stages had great problems . But close to the antenna there was a shadow effect that could not pickup the station at all an antenna aiming problem . It was so pronounced that it changed from nothing to 1.5 volt in a distance of 200 meters. External effects those that one tries to remove when measuring any type of device.
 
Ed, I don't have the energy right now but I distinctly remember you stating that someone left the wire coiled and his measurements were "all wrong" implying that uncoiled would have been OK.
That certainly would be one conclusion, but as even the idiot involved recognized the results were wrong I don't think he got that far.

I do know my measurement rig uses a cheap electret mic capsule to a wireless belt pack system that has no compander. Less dynamic range but matches my wired measurement system to =/- .8 db or so. But the frequency range of interest is only 50 to 10,000 hertz.
 
Thanks Scott and John!

So, it looks that buying Jfet switches for low(ish) noise applications COULD BE a little bit of lottery, since it is not known what they do with the fabs at the factory. Am I correct?

John, 2SK170's are fine and I have plenty of them, I'm just researching what else is still available out there.

I can't find PF5101 DS so can't tell if it is actually same die. ..

Here's one from 1977
 

Attachments

  • PF_NF_5101_3.pdf
    37.3 KB · Views: 70
Thanks Scott and John!

So, it looks that buying Jfet switches for low(ish) noise applications COULD BE a little bit of lottery, since it is not known what they do with the fabs at the factory. Am I correct?

John, 2SK170's are fine and I have plenty of them, I'm just researching what else is still available out there.



Here's one from 1977

Well from that the Vp range is way different than the J111 so there is no way this is the same die.

John simply ignores several years of posts and circuits where folks have had great success with the BF862 which out performs the 2SK170 in almost every way at the price they are not selected for noise but you can hardly find any that are not 1nV at 5mA and 100Hz. Nothing is allowed to change from 40yr. ago.
 
Status
Not open for further replies.