|
Vintage Test Gear and Workshop Equipment For discussions about vintage test gear and workshop equipment such as coil winders. |
|
Thread Tools |
16th Jun 2020, 12:42 am | #21 |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Hi-Z input probe by Bob Pease.
For the purposes of investigation & experiment, I built a slightly modified version: 'modified' on account of available devices in stock. The FETs are BF256 (VHF types); the NPN transistor is a BFY90; the PNP is a 2N4261: both have a high Ft into the UHF region. The gate R for the first FET is a 33 M Ohm metal oxide type. Ample decoupling was installed to ensure stability - which was verified on my spectrum analyzer to 1 GHz. A photo is attached, q.v.
I attempted some measurements. Signal source: HP 8640B; load monitor Tek. 2465 'scope or Advantest TR1432 spectrum analyser. 1. Bandwidth, gain, slew and load O/P capacity. The freq. response was flat to within 1 dB over the range < 1 MHz to > 100 MHz. The voltage gain was zero (as expected, of course). The maximum input was approx. 250 mV rms. Greater inputs than that gave distorted O/P sine-waves, probably on account of slew rate limitation, since as the freq. was increased it was progressively necessary to reduce the amplitude of the input signal to reduce that distortion. The 'offset adjust' pre-set worked as expected, enabling the O/P to be centred on 0v. This amp. was 'happy' to deliver O/P voltages to a load between many kilohms to 50 Ohms - which I found to be somewhat surprising. The connection to the 50 Ohm load (selectable within the Tek. 'scope) was via about 1 metre of 50 ohm co-ax. 2. Input impedance assessment: capacitance and resistance. Using my R & S capacitance meter, I measured the 'no power applied' input capacitance: 3.5 pF. Then I attempted to measure the 'power-on' input capacitance at 1.3 MHz. This was done by inserting a small variable capacitor in a screened box between the sig. gen. and the amplifier's input. When the amplifier's O/P was half of what it was before that variable cap. was added, the value of that cap. was then measured: 7 pF. I account for the difference to be the 3.5 pF I/P cap'y. (as measured) plus 3.5 pF for 'strays'. Using the same approach, I replaced the variable cap'r. with a resistor: 43 k-Ohms reduced the O/P by half. And that puzzles me since at 1.3 MHz, 3.5 pF provides a reactance of 35 k-Ohms; the input R of this amp. is predominantly the 30 M Ohm R, no? Something there doesn't 'add up' . Perhaps my maths. isn't what it used to be! Later, (when time permits), I may try another 'variant' build based on the U440 dual N-FET (typical Cgs of 4 pF) and the BFQ 232 / 252 medium power transistors (Ft of 3 GHz), hoping for an improvement in bandwidth, slew rate and max. Vout. I have no immediate need for this circuit - it just looks worthy of a detailed practical analysis for future reference. Al. Last edited by Skywave; 16th Jun 2020 at 12:54 am. |
16th Jun 2020, 1:02 am | #22 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
Looking at the datasheet I think the BF256 will show a higher input capacitance. I would have guessed maybe 2pF when used in this circuit. However, it only takes a few cm of wire and a few strays to add several tenths of a pF. Could your PCB standoff posts be made with a high dielectric material?
The BF256 and the 2N5485 are both process 50 JFETs but I think the 2N5485 will show lower input capacitance in this circuit because it has lower Cgd capacitance for one thing. It might be worth trying SMD versions of the 2N5485 JFETs if you want to chase the capacitance right down.
__________________
Regards, Jeremy G0HZU |
16th Jun 2020, 1:38 am | #23 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
Have you tried interfacing to a 1Mohm scope input with a square wave at the input? I can't see how the output interface will be able to prevent overshoot/ringing on the waveform if a short run or (say) RG58 coax is used. Things might be different with different transistors but the 2N3906 isn't going to be able to act as a 50 ohm line driver.
That's the reason I asked about the scope interface in a previous post. Has anyone tried looking at fast edges with this probe when connecting to a scope via a short run of coax? I'd expect to see lots of ringing. I'd also expect to see distortion on a sine wave if the amplitude was several Vpkpk at several MHz. Note that this should be done without a 50R termination at the scope end. Just feed the coax into the 1Mohm scope input.
__________________
Regards, Jeremy G0HZU |
16th Jun 2020, 2:27 am | #24 | |
Nonode
Join Date: Oct 2010
Location: Tintinara, South Australia, Australia
Posts: 2,324
|
Re: Hi-Z input probe by Bob Pease.
Quote:
These are his Pease Porridge columns from Electronic Design. Anyway, in the 3rd pdf of the series is an article on stray capacitance including calculating it for FR4, constructing very low pF caps and various other things of interest related to low capacitance effects. The original is from Electronic Design January 22 1996 |
|
16th Jun 2020, 7:38 am | #25 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,939
|
Re: Hi-Z input probe by Bob Pease.
Tektronix, who know a thing or two about probes in general, and the P6201 FET probe in particular. That is a 900MHz bandwidth probe with 100k input impedance shunted by 3pF. That can be reduced by using one of the attenuator heads which provide 1Mohm shunted by ~1.5pF.
The manual, including circuit description, calibration and schematics is here http://w140.com/tek_p6201.pdf . Of course layout, board material and thickness is critical, and they have used various FET's over the long life of this probe, most recently the MMBF4416 surface mount JFET, a process 50 device. Farnell say "no longer manufactured", which is nonsense. ON Semi say it is a current device https://www.onsemi.com/products/disc...jfets/mmbf4416 , and Mouser have 26,609 in stock. Craig |
16th Jun 2020, 10:10 am | #26 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,797
|
Re: Hi-Z input probe by Bob Pease.
2N4416 is quite a useful RF device. and its appeared since in many varieties of packages. It's one of the fairly high IDSS group of JFETS.
I had a prototype receiver by Marconi which used high IDSS JFETs more associated with analogue switching... 2N4391 as low impedance RF amplifiers. The Tek probes show the sort of capacitances you're stuck with, even in the attenuator. Lower C can be dome by bootstrapping and neutralisation, but you get even deeper into concerns over general stability. Pulse and step responses can be both illuminating and depressing! David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
16th Jun 2020, 10:57 am | #27 | |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Hi-Z input probe by Bob Pease.
Quote:
Al. |
|
16th Jun 2020, 11:04 am | #28 | |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Hi-Z input probe by Bob Pease.
Quote:
Data I have obtained for various FETs, Cgs: 2N4416: 4 pF max., typically 2.2pF. U440: 3 pF. BF256: 4 pF. 2N5486: 5 pF. BFJ309@ 5 pF. Al. |
|
16th Jun 2020, 9:26 pm | #29 | |
Dekatron
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,385
|
Re: Hi-Z input probe by Bob Pease.
Quote:
Colin |
|
17th Jun 2020, 1:53 pm | #30 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
Quote:
If the purpose of the probe is to connect to a test point on a PCB whilst adding very little capacitance then I can suggest ways to test this. I have 10MHz and 70MHz RF probe test jigs here that might be able to test the probe in a real world application. I think it might be better to make a lower frequency version of these jigs in this case. It might be unrealistic to test this probe up at 10MHz. Both my probe jigs are designed to be used with a VNA and the VNA sweeps through a lightly coupled resonator on the jig. Next to the resonator is a small and flat gimmick capacitor 'plate' made from a special ceramic substrate that is highly stable in temperature and it is also very stable mechanically. The idea is to put the probe tip on the capacitor plate and then ground the probe nearby. Any shunt capacitance will tune the resonator sideways. Any probe losses due to probe Rp will cause extra insertion loss. The main aim of these jigs was to prove the performance of Schottky diode probes. By varying the VNA source power any variation in detector capacitance or detector Rp due to drive level could be shown by a change on the VNA. Because the resonator steps up the test voltage quite a lot I recall that this jig can test an RF probe up to 20Vpkpk at the resonator if the VNA source power is raised to +20dBm. My old HP8714B VNA had a high source power option fitted and also an attenuator and this allows the source to be varied across something like an 80dB range. I think this setup can just about detect 0.3pF loading on the 10MHz version. I've tested it on the simulated version and also on the real jig and the results agree.
__________________
Regards, Jeremy G0HZU |
|
18th Jun 2020, 8:56 pm | #31 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
I've got hold of some 'new old stock' 2N5485 JFETs and these are still in the Farnell packaging. Unfortunately they are the SMD version so this will make it harder (very fiddly!) for me to play with them in this circuit.
So far I have taken a suite of full 2 port s parameters at various dc operating points and I guess the most relevant would be 15Vds at 5mA Id. I've take s parameters in both common drain and common source. My s parameter test fixture only contains the JFET and this means that all bias tee effects are calibrated out. I also operate the VNA at reduced drive level to maintain the small signal condition. Since doing this I've had an initial play on a simulator using the s2p models exported from the VNA. The results do agree with the linear (process 50) JFET models I have and the non linear JFET models I use with Genesys and it predicts the input capacitance of the probe will be just under 1pF when used with my newly generated s2p models. However, I suspect that the long skinny probe tips used on the original circuit will increase this to just over 1pF. I'll have a go at making the probe over the weekend and I'll try and use a point to point layout in the critical area near Q1. The results for the input capacitance are going to be the same as the simulation because the small signal models the VNA produces are usually very accurate. The uncertainty/accuracy does suffer a bit below about 3MHz when measuring capacitances in the 1pF region but it does do a lot better than many other VNA models. It can usually measure an 0603 0.3pF microwave capacitor quite well down to a couple of MHz. Below this the data gets noisy as the reactance is absolutely huge and the reflection coefficient is very close to 1. The angle will be something like -0.02degrees!
__________________
Regards, Jeremy G0HZU |
18th Jun 2020, 10:28 pm | #32 | |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Hi-Z input probe by Bob Pease.
Quote:
Al. |
|
18th Jun 2020, 10:35 pm | #33 |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Hi-Z input probe by Bob Pease.
Update. I replaced the two FETs with 2N4416 FETs and made various measurements. The general differences were insignificant: all were well inside the relevant regions of 'measurement error'.
Al. |
18th Jun 2020, 10:51 pm | #34 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
You might get better waveform edges with the faster transistors but I think the original circuit will struggle here. Give me a few days and I'll build the original probe myself.
Quote:
I've been playing with the newly generated s2p files of the 2N5485 this evening and I uploaded a short video to youtube. This video doesn't show the Bob Pease circuit because it would get a lot of negative comments if it showed a higher input capacitance than he describes in his notes. However, it does show the benefits of bootstrapping at the drain. The 10k resistor in the source makes a pretty reasonable current source as well. This video shows just how much the input capacitance of a JFET can change depending on the components around it. It isn't a good idea to just look at the datasheet and assume it will be at least as high as the Cgs capacitance for example. The s2p models suggest that the input capacitance can get down to around 0.25pF with bootstrapping via a BJT. https://www.youtube.com/watch?v=7A6k...ature=youtu.be Both the JFET and the BJT s2p models are new but the results look about right to me. Sorry, the video has no sound as I don't have a microphone here and I don't think my old version of Camstudio records audio properly with my OS anyway.
__________________
Regards, Jeremy G0HZU |
|
18th Jun 2020, 11:41 pm | #35 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
If it helps add any credibility to my VNA and component/s2p modelling skills I had a rummage through an old version of Genesys and found some s2p models of the same two transistors that came bundled with the software. The manufacturer's model of the MPSH10 BJT was produced by Philips using a VNA about 30 years ago at the same dc operating point of 10Vce and 10mA Ic.
The JFET s2p model bundled in Genesys is a generic process 50 JFET model and it isn't clear what the dc operating point is. However, if you look at the word doc below it shows a screenshot comparison between both my new models and the 30 year old models from Genesys. Both agree very closely, within about 0.06pF at 10MHz. Their data starts at 10MHz so I've had to start the comparison sweep at 10MHz rather than 3MHz. The agreement for the basic bootstrap circuit is very good indeed! I'm using a very good Agilent lab VNA and calibration system here so I should get very good agreement with the manufacturer's original models. Of course, I also have to make sure I don't make any mistakes in the overall process. I've been doing this stuff for over 30 years but it is very easy to make an error as setting up an s parameter test fixture (with external bias tees) is a long and complicated process.
__________________
Regards, Jeremy G0HZU |
18th Jun 2020, 11:56 pm | #36 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
Note that my test results start at 3MHz. My VNA isn't the right tool to use when it comes to measuring a low capacitance like this at lower frequencies. I could try and rig something with an Analog Discovery 2. I'd have to build the probe and then try and measure the input impedance using the AD2. I haven't used mine in ages so I'm a bit rusty with it...
See below for a quick and dirty simulation with a Spice simulator for a small square wave drive into the probe. I used the On Semi manufacturer's model for the 2N3906 and this is what it shows for in vs out for a 28" length of RG58 coax feeding into an unterminated scope. The result doesn't look bad but it doesn't look good either. I'm not sure how representative the 2N3906 Spice model is when subjected to a harsh test like this so I'm not particularly confident that the real result with the real (original) probe circuit will be the same as this.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 19th Jun 2020 at 12:24 am. |
19th Jun 2020, 11:18 am | #37 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
Quote:
One critical design requirement will be to achieve low input capacitance whilst avoiding any negative resistance anywhere. The Tek probe circuit shows that they use a couple of parallel resistors at the input that act as negative resistance stoppers. The 4.7pF cap between gate and source also helps manage the negative resistance. My simulation does show that the negative resistance gets cancelled quite effectively. However, the really clever stuff will involve the interface to the following PNP BJT. This device will contribute some shunt capacitance and some negative resistance at the source of the JFET and this can be exploited to improve the Rp performance of the probe across the VHF region. I've had a fairly lame attempt at achieving this by adding just a tiny amount of extra shunt capacitance at the end of the video. This helps optimise the design by massaging in some negative resistance to increase Rp. Obviously this can't be overdone or there is a risk of Rp turning negative overall.. The engineers at Tek will be able to improve on this and they will get a better Rp vs frequency profile than my crude model can manage. My model indicated about 3pF Cp and a fairly impressive Rp out through VHF and beyond. I captured the analysis in another quick youtube video. Sorry again for the lack of sound. I'll try and find a microphone at some point... https://www.youtube.com/watch?v=kLJV...ature=youtu.be In summary, this appears to be a professionally designed probe with no hidden gotchas in the form of negative resistance.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 19th Jun 2020 at 11:24 am. |
|
19th Jun 2020, 12:39 pm | #38 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Hi-Z input probe by Bob Pease.
I added just a few nH to the Tek 6201 probe model for the tip inductance and some package inductance for the components and then arranged the Rp and Xp plot scaling to be similar to the plot in the Tek 6201 probe manual.
You can see in the image below that my VNA derived s2p model of a process 50 JFET gave very similar results for Rp and Xp from HF through to 1GHz. You can also see that my model is starting to misbehave above 900MHz as the Xp curve is rapidly turning upwards.
__________________
Regards, Jeremy G0HZU |
19th Jun 2020, 12:54 pm | #39 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,939
|
Re: Hi-Z input probe by Bob Pease.
I'd say that was excellent correspondence between Tek's measurement and your modelling.
|
19th Jun 2020, 1:02 pm | #40 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Hi-Z input probe by Bob Pease.
I saw the addition of that capacitor as a way of swamping (to some extent) the inevitable sample-to-sample variations in Cgs, so at least they can deal with a more consistent capacitance. Know thy enemy
|