John Curl's Blowtorch preamplifier part II

[ikoflexer]

Simon, have you described anywhere how you make these measurements? If not, will you?

[Joachim Gerhard]

Yes, the Röderstein are KP1832. I have them in 1.5nF-1200V, 10nF - 1500V, 18nF - 2000V.
They made also MKP1842 and MKP1839 with higher values that are also very good.

[john curl]

Germans usually make better than usual quality stuff. I have a lot of Siemens polystyrene from the past, old Roderstein resistors, etc. I tend to use RELIABLE caps, made in USA, for serious caps. I should not tell you this, as you are a potential competitor, but we are also colleagues.

[bear]

My somewhat ignorant impression is/was that the so called "non-inductive" wirewounds are not reactive up to some relatively low frequency when we are into the RF range... is this correct?

_-_-bear

[john curl]

They are pretty good, but with feedback amplifiers, I would expect some compromise in stability.

[1audio]

I think I have a pretty good understanding of Simon's measurement technique and its pros and cons. I do have a concern that two slightly nonlinear devices that are almost perfectly matched will cancel their nonlinearities in the middle. Also given that the distortions are in the parts per billion range and the setup will amplify in a nonlinear way any impedance differences at higher frequencies it would need the parts to be similar to a very high degree. Even the balancing (lead compensator) network will impact the overall performance at this level of resolution. I have played around with this type of measurement for DC calibration if resistive divider networks to the ppm level and its very sensitive to setup even at DC. At AC there are issues about stray C's and L's not to mention the balance of the differential amps input circuitry. Attempting to duplicate John Curl's cap measurements was quite the challenge. For an in depth discussion read this http://assets.fluke.com/manuals/720A____imeng0200.pdf

I sent Simon a set of 20 matched (seem to be to about .01%) pairs of 10K Vishay Z102 resistors. These all measure in the -165 dB + range on my CLT 1. It will be interesting to see how they work in Simon's setup.

As for wirewound Resistors and high frequencies- the HP 400 series of AC voltmeters (1% accuracy to 4 MHz) used wirewound resistors on mica cards on the input divider network. Its all relatively high impedance so stray C is a big issue but the resistors were able to meet those very stringent requirements, and I would not write them off today either.

[simon7000]

Demian,

Got em. 20 dual 10Ks, 5 3.33K, 2 1K rect, 2 10K axial, 1 1KWW, and 1 1K caddock.

As my current setup is for a 10K bridge with the DUT upper left and lower right I will have teat the duals as singles, The 3.33K I will use 3 in series against a reference, the 1Ks will have to wait for a new setup. I have so far gone through my "Audio", military, commercial MF, and Carbon film samples. I should finish the carbon comps tommorrow and start on the surface mounts, so I should have your data by Friday for the easy ones!

[Joachim Gerhard]

We will use RELCAP capacitors too in our American made Sonics speakers.
I talked to the designer at CES and he told me that he buyes the raw plastic foils in Germany so it is a kind of German-American friendship project.

[1audio]

Simon,
I somehow though you were using 1K as your target. No need to check 40 resistors unless you want some statistics on s102 Vishay chips. You could also set up two of the Vishays as a bridge and see what you find as a balance.

If work is light this week I'll build a similar fixture and see what I learn.

[simon7000]

Originally Posted by 1audio

Simon,
I somehow though you were using 1K as your target. No need to check 40 resistors unless you want some statistics on s102 Vishay chips. You could also set up two of the Vishays as a bridge and see what you find as a balance.

If work is light this week I'll build a similar fixture and see what I learn.

Never mind, I am running 1K's at the moment. Should be able to test them all. I will run the vishay 10K's all by them selves and make a 1k set for comparison.