Hantek - Tekway - DSO hack - get 200MHz bw for free

[tinhead]

How to make better external 50ohm temination if scope have not internal 50ohm impedance matched inputs?

with "active" terminator, like e.g. this project :

http://www.mikrocontroller.net/topic/204806

The guy who working on this project designed already nice active probes:
http://welecw2000a.sourceforge.net/docs/Hardware/Aktiver_Tastkopf_mit_OPA659.pdf

... if scope have not internal 50ohm impedance matched inputs?

attached how a 50R termination looks inside Rigol and ATTEN DSOs, i doubt this constrution is much better than
external solution. Everything else is the same as in HanTekway.

[jahonen]

I think that it matters what kind of capacitance it is, the amount itself is probably not very important. Lumped capacitance is a bad thing, distributed capacitance is not (think about coaxial, DC inductance or capacitance does not affect AC performance, only resistance and dielectric loss).

Problem with the external terminator is that there is a stub after the terminator causing reflections. The termination resistor should be the very last thing on the transmission line. Internal termination is naturally superior to an external one as it eliminates the stub (depending on the internal construction, of course).

Regards,
Janne

[rf-loop]

@rf-loop: I checked the Tek 2465 schematics, 50 ohm coupling only consists in switching in a 50 ohm resistor in parallel to the 1 M resistor. The only additional trick is to put a thermal sensor on the 50 ohm resistor to check for channel overload.

Anyway, 50 ohm // 1 Mohm = 49.9975 ohm 

This schematic do not show whole truth. As we know all wires have inductance, capacitance and resistance. Whole complex is build so that it is "matched" 50ohm termination. (of course it is not very accurate if we go more high frequencies.)

Do you know looking this schematics what is example inductance in this line where is  "50" ohm reistor.
I have many times repair and service this input attenuator. If think its RF charasteritics need know exactly all dimensions etc.

I have not in home any VNA but still after I have looked inside I "believe" that there have been enough good RF designer who have designed it.

But this Atten looks strange. Nothing but 50 ohm resistor... well yes it is 50ohm if think DC. But how is this kind of system with RF.

Also I have deeply looked old HP input attenuator including same 50 ohm terminator. (it is "hybrid" package where also relays are on the thick layer/ceramic hybrid what include  DC/AC(RF) attenuators and matched ternminator. I have no any document about this story but I trust one my friend who tell that he have looked it with VNA and it is really well designed to over 400MHz. If there is not any reactive compensation for this input capacitance it can not be 50 om matched terminator for RF, only for DC. So I really believe that TEK 2465 input attenuator module is much more complex as you see in this service manual schema. (this schema is only indicative, and not all reactive things are shown as component. One strip line is inductance, cpacitance and resistance "network" if we look it as RF signal look it)
example  10nH inductance is "lot of"

I can not see in this schema example this 50ohm resistor inductance and capacitance. So.. but in real world there is lot of these in many place..

So with this Tek input schema we can not tell nearly anything anbout ihow it wok with high frequencies. We need know whole mechanical dimensions. Wires dimensions etc. Around 200-400MHz it is still quite simply... but more high we go... it looks like more complex.)

[tinhead]

@tinhead:
From which model is this picture from? From all the front-end pictures I have seen so far,
I never observed the ability to get a 50 ohm input termination!

this is actually from Rigol CA, however ATTEN CAL looks same, and yes there are no any pictures of both
available public.

Even worse: in most of these scopes, there is no real GND coupling and the Rigol does not even have AC coupling for trigger...

sure there is GND coupling, the AD8370 PWUP pin is responsible for this.

[tinhead]

@tinhead: GND coupling should actually isolate completely the DUT from the front-end! At least, this is what I observed in ALL other scopes before. But of course, this costs one additional relay 

sure, but that's additional 2pF in signal path. From a scope point of view "GND coupling" over PWUP pin
is what it suppose to do, from DUT point of view not, but to be very honest who cares.

[tinhead]

I wouldn't leave this unpowered DSO connected to my DUT, if I were you

why ? I don't see any issues (at least hw0), unpowered DSO from DC point of view have 1.2M to GND, nothing else (att is off, PhotoMOS off, 330pF blocking DC to JFET). From AC point of view i will not kill that fast the JFET, especialy that
i'm always using x10 passive or active probes (except DSO testing where of course shit can happens
when i'm too lazy to power on DSO first).

[rf-loop]

I completely agree with you rf-loop, of course, a schematic doesn't tell you the whole story

EEs at Tek, HP and LeCroy really knew their job: even 20 years later, these scopes are still nice pieces of jewelry!

To give you an idea of what these scopes are made of, please look at this very interesting April 1986 issue of the HP Journal, almost dedicated to "Digitizing Oscilloscopes", and to the HP5411OD in particular.

But we should compare what is comparable: these were x$10,000 scopes, vs. $300 ones! These old scopes were built around hybrid circuits with coupled microstrip lines and delay lines, our cheap DSOs are made up of 6-layer FR4 with stock components...

So for sure, you can't expect the same performance from both. Even if progress has made things cheaper, RF rules have not changed 

Yes.

Also I did not mean anything for comparing these. (exept that today these old work horses are still very good for real working. Working value is good if look these old work horses price today)

My only menaing is note that this unmatching really affect and we need always keep it n mind. As I or we tell that xxx scope have -3dB bandwidth example 300MHz it is more or less true. We only know whole test setting "system" BW. This very simple example explain that even with quite "good" test system there may be lot of error. Using example TEk2465 with internal or external terminator give big difference. This do not proof anything about example Hantek. But it give some data for thinking.

Some times ago I ask here something like: What is Tektronix "input normalizer" box what need use for Tek calibrations. Only for adjusting input capacitor. But it is more...and it may give some wink also how to "delete" scope input capacitance and make better match. (it is passive and it make attenuation but 

@tinhead   
thank you. This link to  talking about matching was nice. 
Some day I will make small "matching box"

[rf-loop]

To give you an idea of what these scopes are made of, please look at this very interesting April 1986 issue of the HP Journal, almost dedicated to "Digitizing Oscilloscopes", and to the HP5411OD in particular.

This is nice article I have read many many years ago. (and now look agen.. HP journal's have been very nice and lot of good articles) I have 54111, (and of course many others but today only some becouse I have sell them out) but today its 1'st power what serve analog and digital power and CRT is bad and I have not get spare part. I get this scope so that  channel front end attenuator unit was bad. But I repair it. After then later power unit get smoke out. (switch controller IC is bad and this special controller IC is really hard to find...)

[patb]

Thanks guys for a lot of info in this thread (esp. tinhead). It helped me to choose my first scope. I bought Hantek 200 MHz version (for almost the same price as 100Mhz). I was playing with my new toy for a while and discovered a little strange behavior. When I use only CH1 and cross 200ns time base boundary it switches probably to interlaced mode (my guess, I hear relay click) and I get a little bit different readings. See attached images. The change is subtle but noticable. Is this a normal behavior?

[tinhead]

When I use only CH1 and cross 200ns time base boundary it switches probably to interlaced mode (my guess, I hear relay click) and I get a little bit different readings. See attached images. The change is subtle but noticable. Is this a normal behavior?

from 40s to 400ns the DSO is using always 4 ADCs per channel, no matter how many channels enabled
(so max 400MSs sampling rate each channel).

From 200ns to 2ns the DSO is using 8 ADCs when single channel enabled (so 800MSs and with 8/4/2ns timebase 1GSs)
and 4 ADCs when both channels enabled (so 400MSs and with 8/4/2ns timebase 500MSs each channel).

As from your observation, sure by using only one channel the DSO (when switching from 400ns to 200ns)
is switching the sample rate from 400MSs to 800MSs, therefore you will see small difference.

[patb]

Thank you tinhead for explanation. Actually, I observed that little drop of readings only at 800/400 ns. Not a big deal, but it is always good to know what's going on inside.

Cheers!

[pionersurg]

tekway dst1102b  FW 2.06.3 (110531.1) HW 10070x555583e8
I want only load english UI.
If I correcly understand:
1.download 2.06.3_110531.1_FullCSVSave_And_HelpFileCorrected.zip and unzip to usb stick
2.connect to UART (UA0) using RS232 to USB converter (i use pl2303), settings 115200 8n1.
I connect pin2 RX to TX pin on board, pin3 RX to RX pin on board, pin5 GND to GND pin on board.
When power ON terminal emulation program receive many symbols, but no any command prompt. I try different speed setting COM port.

[rf-loop]

Do NOT connect RS232 standard port to UART port!

You need example USB TTL 3,3V serial adapter

Just like this:
http://cgi.ebay.com/USB-TTL-Converter-Module-buildin-in-CP2102-NEW-/230648740143?pt=LH_DefaultDomain_0&hash=item35b3bc792f

Maybe you have right adaper.

 (and look carefully how connector is on the HW7!!)

AFAIK also tekway have 15200,8n1  First chek your wires and connectors or adapter.

If it do not work your adapter or connection is fail somehow.
Check carefully that you are relly connected to UART0
on the board (HW7) reading  from front panel end you use pins 2,3 and 4.
UARTplace and direction  is changed after HW5, it is nearly CH1 box.  J801 pin 1 is nearest to front panel. Pin2 is GND

Then problem may be EMI (make good shielded and look also that scope and computer are so that there are not any ground currents.)

As you start your scope it send lot of text to terminal.
If you see there errors. Find problem before you go on with ctrlC

Also my recommendation is. Connect firts GND! Example connector where GND contact littlebit before other pins. and then signals.

UART is directly to ARM. There is not any kind of buffering and protection. It is very "fragile"

[snoopy]

Hello everyone,

Just joined up.

I came across this thread looking for info on the Hantek DSO5202B which I subsequently purchased from an ebay seller.

The CRO works really well but I have struck a few issues with this CRO which I wonder if other people have found. I have sent this info off to Hantek but have not heard back from them. The firmware installed is later than what is on the website.

For instance, AC coupling of the input channels inserts a high pass filter that peaks at 100 Hz and rolls off below that. So when you display a 100 Hz sinewave and switch from DC to AC coupling the signal amplitude actually increases !! Ideally the AC coupling should be flat down to 10Hz and roll off below that without any peaking otherwise this is causing an error in the readings.

Also frequency measurement on the bottom of the screen does not correlate with the menu because the bottom screen measurement decimal places are always zero. ie frequency shown in measurement menu is 105.3Hz and frequency shown at the bottom of the screen is 105.00 Hz.

regards

[rf-loop]

Hello everyone,

Just joined up.

I came across this thread looking for info on the Hantek DSO5202B which I subsequently purchased from an ebay seller.

The CRO works really well but I have struck a few issues with this CRO which I wonder if other people have found. I have sent this info off to Hantek but have not heard back from them. The firmware installed is later than what is on the website.

For instance, AC coupling of the input channels inserts a high pass filter that peaks at 100 Hz and rolls off below that. So when you display a 100 Hz sinewave and switch from DC to AC coupling the signal amplitude actually increases !! Ideally the AC coupling should be flat down to 10Hz and roll off below that without any peaking otherwise this is causing an error in the readings.

Also frequency measurement on the bottom of the screen does not correlate with the menu because the bottom screen measurement decimal places are always zero. ie frequency shown in measurement menu is 105.3Hz and frequency shown at the bottom of the screen is 105.00 Hz.

regards

I test with HW7 DSO5102B (in this case very same as 5202B)

AC mode (Why?)

If compare to 1kHz sine signal level (set in calculations as 0dB reference)

Measured directly without probe.
Including old Wavetek flatness errors (yes but this my Wavetek is carefully adjusted so that it is surprising flat level - error is significantly below +-0.1dB in this area)
CH1, AC, 500mV/div
signal level 6div p-p @ 0dB

-6dB 10Hz
-3dB 17Hz
0dB 41Hz
+1dB 80Hz
0dB 350Hz - 1kHz

----------

About frequency counter.

Measure menu freq are _calculated_ from ADC captured data.
With low freq it may sometimes be better and more accurate than HW counter. Specially if use averaging.

HW counter work totally different. It counts events in trig line. It is event counter. It counts how many events in time window.
Becouse it is not as modern real timeintervall counters it resolution is highly dsependent about time window setting for counting. How long is gate open window what time it use for counting these pulses. (yes it is just pulse counter. Totally different as modern frequency calculating time interval counters as something like cheap HP53131 or similars. They can solve 9-10 or more decimals in one second.)
 
How many seconds need count if you want count 50Hz signal frequency with 1 or 0.1mHz resolution (mHz is milli Hz) if use simple pulsecounter type frequency counter.  I think no one want waiting.

But this win clearly measure menu freq display if we go more high frequencies.
If you have 999999Hz you have one Hz resolution and counting time is something like 1s.
It give lot of more accuracy with HF signals.
As understand this you can thing which one you look in which case. (yes with low freq it is maybe clever if Hantek shut off these zeroes if they are not meaningfull.


Also it can read  in specs (exept that +-30ppm is lie.. maybe they still do not know how to make tests for specs and what are normal "good" rules for  specifications.:
(I will write it +-150ppm over temp area and including aging. If can not get any other info There meybe need write lifetime aging drift becouse it have NO adjustment! )
maybe they need write this littlebit more clear so that peoples undertand better how it works and what are limits.
It is pure pulse counter.  It "blind" open and close gate and count pulses what come as gate is open. gate time is 1s.
(but if there is example 50Hz, why it do not shut off all zeroes after decimal point. FW bug?

User manual (rev.9)specs
Readout Resolution 6 digits
Accuracy (typical)
±30ppm (including all frequency reference errors and ±1 count
errors)

Frequency Range AC coupled, from 4Hz minimum to rated bandwidth

Signal Source
Pulse Width or Edge Trigger modes: all available trigger sources
_The Frequency Counter measures trigger source at all times,
including when the oscilloscope acquisition pauses due to changes
in the run status, or acquisition of a single shot event has
completed._

Pulse Width Trigger mode: The oscilloscope counts pulses of
significant magnitude inside the _1s measurement window_ that
qualify as triggerable events, such as narrow pulses in a PWM
pulse train if set to < mode and the width is set to a relatively small
time.

Edge Trigger mode: The oscilloscope counts all edges of sufficient
magnitude and correct polarity.

Video Trigger mode: The Frequency Counter does not work

[snoopy]

Hello everyone,

Just joined up.

I came across this thread looking for info on the Hantek DSO5202B which I subsequently purchased from an ebay seller.

The CRO works really well but I have struck a few issues with this CRO which I wonder if other people have found. I have sent this info off to Hantek but have not heard back from them. The firmware installed is later than what is on the website.

For instance, AC coupling of the input channels inserts a high pass filter that peaks at 100 Hz and rolls off below that. So when you display a 100 Hz sinewave and switch from DC to AC coupling the signal amplitude actually increases !! Ideally the AC coupling should be flat down to 10Hz and roll off below that without any peaking otherwise this is causing an error in the readings.

Also frequency measurement on the bottom of the screen does not correlate with the menu because the bottom screen measurement decimal places are always zero. ie frequency shown in measurement menu is 105.3Hz and frequency shown at the bottom of the screen is 105.00 Hz.

regards

I test with HW7 DSO5102B (in this case very same as 5202B)

AC mode (Why?)

If compare to 1kHz sine signal level (set in calculations as 0dB reference)

Measured directly without probe.
Including old Wavetek flatness errors (yes but this my Wavetek is carefully adjusted so that it is surprising flat level - error is significantly below +-0.1dB in this area)
CH1, AC, 500mV/div
signal level 6div p-p @ 0dB

-6dB 10Hz
-3dB 17Hz
0dB 41Hz
+1dB 80Hz
0dB 350Hz - 1kHz

----------

About frequency counter.

Measure menu freq are _calculated_ from ADC captured data.
With low freq it may sometimes be better and more accurate than HW counter. Specially if use averaging.

HW counter work totally different. It counts events in trig line. It is event counter. It counts how many events in time window.
Becouse it is not as modern real timeintervall counters it resolution is highly dsependent about time window setting for counting. How long is gate open window what time it use for counting these pulses. (yes it is just pulse counter. Totally different as modern frequency calculating time interval counters as something like cheap HP53131 or similars. They can solve 9-10 or more decimals in one second.)

thanks for the reply

+1dB 80Hz

so I can see that your's is not flat around 80Hz. Mine peaks right at 100Hz. This is not good and can be very misleading if you are not aware of it Hopefully they will be able to fix it up with a firmware upgrade.

And why use AC coupling ? Simply because you may have a small AC signal superimposed on a large DC offset so you want to be able to get rid of the DC component otherwise it is not possible to view the small AC signal with maximum sensitivity.

Regarding the counter at the bottom of the screen I did not know that the two counters were different which explains why they display different results to each other.

regards

[rf-loop]

And why use AC coupling ? Simply because you may have a small AC signal superimposed on a large DC offset so you want to be able to get rid of the DC component otherwise it is not possible to view the small AC signal with maximum sensitivity.

regards

Yes, this is mostly nearly only reason.

This is not FW related thing. It is HW. 
Analog front end is simple and cheap. It is just like compromise.
If we look example old expensive HP or Tektronix front ends they have lot of adjustments for corrections and corrections for corrections.. more accurate want, there is more small errors what need compensate with some adjustable or not adjustable circuit.

Specially I do not like this way they do AC/DC in front end. From designer I can ask why? Some reason? Any reason?
Or reason is becouse copymachine did not change it. It is problem with copy, you copy also mistakes or bad's.

Also there are all component tolerances. Tolerances may be sometimes in worst case compination and randomly sometimes best case. (example this peaking freq difference - but in this case you have real 5202B what have some different componet values in front end so it may also affect this difference.)

I do not like this front end.  I can not understand this solution for AC/DC. (or maybe if really think every cent)

I have recommend to some my customer (who want more low cut off frequency) to use external DC block.

I make one short experience.

Scope input DC
Probe 1x
Measured
From 50ohm output terminated with 50ohm  feed trough terminator.

For DC block: 100nF capacitor from terminator center to probe tip.
Response "flat"  to around  20Hz
very slowly it satrt drop. 10Hz 3V-pp have drop to 2,9V
1,5Hz it have drop to 2,14V

Lack of time I did not test with 1uF

Probe 10x
as before, 100nF

in practice, nearly flat down to 1Hz.

This can give also idea how to measure extremely low frequencies over high DC. (1uF, 10uF and even more but be careful if voltage is high... becouse..if there is transient... well: after smoke come out from component, equipment do not work...smoke need put back agen)

In lab it is many times needed. DC block. (for common use it may be total block or sometimes only ground connecting "center" block.)  For probe tip DC block, well, everyone can imagine how to do if need.

In some rare case I have use 100uF polypropylene capacitor with some Tektronix. (some reason need measure extremely low freq's over high DC)

[snoopy]

And why use AC coupling ? Simply because you may have a small AC signal superimposed on a large DC offset so you want to be able to get rid of the DC component otherwise it is not possible to view the small AC signal with maximum sensitivity.

regards

Yes, this is mostly nearly only reason.

This is not FW related thing. It is HW. 
Analog front end is simple and cheap. It is just like compromise.
If we look example old expensive HP or Tektronix front ends they have lot of adjustments for corrections and corrections for corrections.. more accurate want, there is more small errors what need compensate with some adjustable or not adjustable circuit.

Specially I do not like this way they do AC/DC in front end. From designer I can ask why? Some reason? Any reason?
Or reason is becouse copymachine did not change it. It is problem with copy, you copy also mistakes or bad's.

Also there are all component tolerances. Tolerances may be sometimes in worst case compination and randomly sometimes best case. (example this peaking freq difference - but in this case you have real 5202B what have some different componet values in front end so it may also affect this difference.)

I do not like this front end.  I can not understand this solution for AC/DC. (or maybe if really think every cent)

I have recommend to some my customer (who want more low cut off frequency) to use external DC block.

I make one short experience.

Scope input DC
Probe 1x
Measured
From 50ohm output terminated with 50ohm  feed trough terminator.

For DC block: 100nF capacitor from terminator center to probe tip.
Response "flat"  to around  20Hz
very slowly it satrt drop. 10Hz 3V-pp have drop to 2,9V
1,5Hz it have drop to 2,14V

Lack of time I did not test with 1uF

Probe 10x
as before, 100nF

in practice, nearly flat down to 1Hz.

This can give also idea how to measure extremely low frequencies over high DC. (1uF, 10uF and even more but be careful if voltage is high... becouse..if there is transient... well: after smoke come out from component, equipment do not work...smoke need put back agen)

In lab it is many times needed. DC block. (for common use it may be total block or sometimes only ground connecting "center" block.)  For probe tip DC block, well, everyone can imagine how to do if need.

In some rare case I have use 100uF polypropylene capacitor with some Tektronix. (some reason need measure extremely low freq's over high DC)

Yes, traditionally a DC blocking capacitor would be used and a switch or relay used to short it out for DC coupling. It appears they don't do this in this CRO for some odd reason, maybe to save cost on a relay etc.

A more practical solution would be an external BNC probe adapter with a capacitor and a switch that shorts it out etc.

The whole implementation with this CRO is really quite silly IMHO.

regards

[rf-loop]

Yes.

After I first time see this schema I was surpriced how they have solve AC/DC coupling and also surprised GND "solution" (=nothing)

This frontend principle is well known and I know some, example Tektronix, have use it. But not just exactly same.

This application for this principle is not best of.  Specially just flatness (not only low end but also overall flatness is only "acceptable/good" but far away from perfect (but, think price and other values in this scope). Also I'm not sure how it works with two paths signal phase shift effects.  I have not analyzed and mesure inside scope how it just product this LF flatness problem.

I do not know any Tektronix scope what have this problem and use same principle in front end. And I have handle, repair and sevice and calibrate tens of Tek's.

But IMHO, this hantek solution is still far away from silly.  First you look price! (What is the factory's price. Not factory selling price)

Of course it can be better - all can.

I can not find easy way how it can do without extra costs. Also there must not be any extra things what need factory adjustments becouse costs.

Other problem is that input capacitance is so big (specially if think >200MHz) that it is not good to rise any pF. (better if find solution to drop it - yes it can markable drop if make separate thick layer ceramic hybrid with special relays... but, after then .. why do expensive good front end ... then ned also do better ADC. Oh well, why then do not buy Agilent. So... this is just good compromise with price and specs. (maybe this 1dB peaking around 100Hz is still not bad "real" problem. It is nearly as "cosmetic".  Specially becouse in this rare case that really ned make accurate low freq amplitude measurements over high DC.  (but, this scope is markets in class where price competition is extremely hard)

(20pF reactance is around 40ohm with 200MHz! (But there is opinions that today input C is littlebit less and if all is ok it is not still bad. (Originally expensive work horse Tektronix 350MHz have 15pF so capacitance is not any problem.)


maybe there is some posibility to adjust this LF/DC amp response better with changing littlebit some componets values but... then maybe some other thing go bad?

[snoopy]

I think this front-end design is not bad, especially for its price and using only stock components.

Regarding the LF loopback, the "corner" point at which it occurs is given by the capacitor in the loop and I don't remember which resistors around. But it occurs at some 100s Hz, and if position is well compensated, should be smooth to HF without too much hiccups. This shouldn't affect the frequency response above that.

In order to understand what's going on, we should put a probe on the AD8370 input (TP2_1), first, then a differential probe between the 2 test points at the LMH6552 output, and see what we get. I suspect that the varicap introduces non-linearities, but I am not sure.

Then what is displayed on the LCD, that's another story... The ADC clock seems to have a lot of jitter, so this is not good for ADC interlacing and may thus produce a lot of distortion, and running without any anti-aliasing filter too is a bad idea, as you may get additional distortion by spectrum folding at the sampling frequency...

do you have the schematic of the front end so I can see what you are talking about ?

regards
david

[snoopy]

I think this front-end design is not bad, especially for its price and using only stock components.

Regarding the LF loopback, the "corner" point at which it occurs is given by the capacitor in the loop and I don't remember which resistors around. But it occurs at some 100s Hz, and if position is well compensated, should be smooth to HF without too much hiccups. This shouldn't affect the frequency response above that.

In order to understand what's going on, we should put a probe on the AD8370 input (TP2_1), first, then a differential probe between the 2 test points at the LMH6552 output, and see what we get. I suspect that the varicap introduces non-linearities, but I am not sure.

Then what is displayed on the LCD, that's another story... The ADC clock seems to have a lot of jitter, so this is not good for ADC interlacing and may thus produce a lot of distortion, and running without any anti-aliasing filter too is a bad idea, as you may get additional distortion by spectrum folding at the sampling frequency...

do you have the schematic of the front end so I can see what you are talking about ?

regards
david

You can find it as an attachment to this tinhead's post.

OK thanks.

I can see the AC/DC relay and DC blocking cap but where is the peaking coming from at 100 Hz ??

Possibly need to simulate it in spice to see what is going on.

regards
david

[snoopy]

I can see the AC/DC relay and DC blocking cap but where is the peaking coming from at 100 Hz ??

Please read again my reply above, when a speak about the corner frequency at some 100s of Hz. It looks like the transition from LF to HF is not well calibrated.

Possibly need to simulate it in spice to see what is going on.

Possible, but not easy... You must make provisions for stray capacitance due to component pins and also to the FR4 PCB (with permeability tolerance of around 50%), and circuit loop inductance and also mutual inductance between circuit loops

When you get a model that is close enough to reality, let me know, I will be very interested, and probably others too!

yes but at very low frequencies surely all of those things are not going to be relevant ?

regards
david

[tinhead]

placeholder

[tinhead]

placeholder ?

[tinhead]

placeholder ?

check next page to see why ..