Hello!
How do I test real jFET's for the value of rdson? And what graphics should I be looking for? I'm talking bout testing on the breadboard, and I would like to do a graphic of current Ids by voltage Vgs, am I right?
A very simple method is to create a divider,
VDC ---> R1 ---> JFET ---> measure voltage across JFET
VDC is known/measured, R1 is known/measured, you can then solve for Ids and rds_on.
You can argue the impedance of the voltmeter is much higher than the JFET, or you can take it into account.
Build the experiment in spice to verify it works. Then compare the experimental results.
I suggest varying R1 in spice to set different voltages on the JFET. Find the point where Vgs is high enough to affect the rds_on value. The experiment needs to keep the voltage across the JFET below that point if you want to trust the results.
Next I suggest measuring it with a multimeter on different ohm scales. The problem with that is you don't know what Vgs is used by the multimeter.
Quote from: Rob Strand on September 25, 2020, 02:42:22 AM
A very simple method is to create a divider,
VDC ---> R1 ---> JFET ---> measure voltage across JFET
VDC is known/measured, R1 is known/measured, you can then solve for Ids and rds_on.
You can argue the impedance of the voltmeter is much higher than the JFET, or you can take it into account.
Build the experiment in spice to verify it works. Then compare the experimental results.
I suggest varying R1 in spice to set different voltages on the JFET. Find the point where Vgs is high enough to affect the rds_on value. The experiment needs to keep the voltage across the JFET below that point if you want to trust the results.
Next I suggest measuring it with a multimeter on different ohm scales. The problem with that is you don't know what Vgs is used by the multimeter.
Nice! Thanks. Based on that I did sims on Spice and went to measure real values on breadboard .
What I did was 9V battery on bread and a total resistance of about 10,3 megaohm. I measured the voltage across the last one and made math to find current. Then I measured the voltage across Drain and Source (source and gate were connected to ground) and got 0.3mV. Did math and foumd that the rds is about 405 ohm if i'm not mistaken.
The only thing is the precision of the multimeter, which does not pass the .1 mV (so I don't know if it was 0.32, or 0.34, or 0.30 mV.. You get the point).
> total resistance of about 10,3 megaohm
So 1uA?
> got 0.3mV
If you look over at my simulated test of simulated JFETs, you see the Rds hardly changes from uA to nearly a mA.
Testing near 100uA (0.1mA, 9V in 90k, or 100k nearest value) gives a voltage like 100mV-500mV, and you do not care about the last 0.1mV.
QuoteThe only thing is the precision of the multimeter, which does not pass the .1 mV (so I don't know if it was 0.32, or 0.34, or 0.30 mV.. You get the point).
If you have poor resolution the measurement error goes up. So it would be wise to raise the test voltage.
There are two types of errors,
- approximation error where you are driving the JFET at a high voltage and the behaviour changes
- rounding error where you are losing digits.
There is a point in the middle where you get the best accuracy.
Currently you are measuring with a rounding error of 30%, which is terrible. You probably want that to be a truncation error of better than 10%, say 1%. That will be a test voltage of 10mV. Like PRR mentioned the you should be able to drive a lot harder than that. Plot Rds_on and what voltage the error is at 1%.
Quote from: savethewhales on September 24, 2020, 08:18:28 PM
How do I test real jFET's for the value of rdson? And what graphics should I be looking for?
look at figures 2 and 3 in this paper :
https://www.vishay.com/docs/70598/70598.pdf (https://www.vishay.com/docs/70598/70598.pdf)
it applies to AC signals ...
---
so, I grabbed a pair of Fairchild 2n5457's this morning and did a poor-man's Vgs(off) and Idss test using a fresh 9v battery and a crappy (1Meg input) DMM
and came up with the following
-1.404/2.96mA and -1.414/2.97mA for Vgs(off) and Idss
---
if we look at the theory : (https://coefs.uncc.edu/dlsharer/files/2012/04/J3a.pdf (https://coefs.uncc.edu/dlsharer/files/2012/04/J3a.pdf))
we see a quadratic equation for the ohmic region
rds(on) is the measure of the slope of the curve corresponding to Vgs=0,
where it crosses the origin ie., at Vds=0 ...
taking the partial derivative of that equation, while setting Vgs and Vds to 0, leads
to the following: gm = -2KVp where K is Idss / Vp^2
giving us rds(on) = Vp / -2Idss
suggesting an rds(on) value of about 238 ohmsAC for both devices, give or take
---
a 1k pot was set to 238r resistance to form a voltage divider between the pot and a DUT ... with an AC test amplitude of 100mVpp derived from an HP 3310 signal generator something very close to 50mVpp across the DUT was observed on a scope ... same transfer on both FETs
tweaking to 326r gave rough 50% transfer ...
I'm glad this topic came up, I have wondered about Rds-ON for JFETs too, and been frustrated that it isn't mentioned in most datasheets. If it's a few ohms, who cares. But several hundred might make a difference somewhere. I'm thinking of a case like a true bypass TS808 clone, compared to an actual Ibanez that would have a 2SK44 or 2SK30 in the audio path. I think those types were selected for the switching because they have fairly low Rds-ON, so it may still be irrelevant. But maybe if the goal is to make a more accurate clone we should be sticking a small resistor in place of the JFET.
I have at least a couple 2SK30s. I get the gist of the test experiment Rob desrcibes above, but I'm not fully clear on how to hook it all up on the breadboard.
> Rds-ON for JFETs too, and been frustrated that it isn't mentioned in most datasheets
JFETs are sold as two semi-arbitrary classes: Amplifier and Switch.
Switch Rds(on) is a key parameter and normally mentioned in the sheet.
Amplifier Rds(on) is nearly irrelevant: if you get near there you are distorting. However for general planning you can figure 1/Gm
(1/Yfs) as a fair approximation of Rds(on).
2SK30 is clearly an Amplifier. If they use it as a Switch it is just to consolidate the BoM.
Quote from: PRR on September 27, 2020, 08:52:02 PM
you can figure 1/Gm
I've seen that mentioned before, in one case a post by you on another forum (GroupDIY?). But I don't see Gm on datasheets either, and don't know how to measure or calculate it. I know it's transconductance and usually expressed in µmhos, but that's it.
Quote(1/Yfs)
I have no idea what Yfs is.
Quote from: willienillie on September 27, 2020, 09:00:59 PM...I don't see Gm on .... I have no idea what Yfs is.
https://datasheet.octopart.com/2SK30A-Toshiba-datasheet-101654.pdf
(https://i.postimg.cc/v43V1d8K/2-SK30-Yfs.gif) (https://postimg.cc/v43V1d8K)
Okay, I guess I was just skimming the Units column looking for µmhos.
mS is milliseconds?
Yes I'm still confused. The minimum Yfs number of 1.2mS would equate to less than 1 ohm (1/1.2) Rds-ON?
Here's a plot of the percentage errors for rds_on and the percentage error for 0.1mV Vgs measurement resolution.
Both y-axis scales are percentage error. I'm assuming the error truncates as this is worse than rounding, in reality we don't know what it is but it's likely to round.
(https://i.postimg.cc/bsS6N7jT/JFET-Rdson-Percentage-Error-2020-09-28.png) (https://postimg.cc/bsS6N7jT)
The optimum test voltage is about Vgs = 25mV with an error of +/- 0.4%.
Testing at Vgs = 60mV the error goes up to 1.0%.
You can also take many measurements at different voltages an extrapolate the rds_on back to Vgs = 0.
Quote
mS is
It's m = milli = 1e-3 and S = Siemens = 1/ ohms. Siemens is the units for conductance and conductance is 1 / ohms.
Conductance = 1 / resistance, symbol is G
Admittance = 1 / impedance, symbol is Y ; that's where the Y comes from.
1.2mS = 1.2 e-3 S = 0.0012 S
rds_on = 1/Yfs = 1/0.0012 = 833 ohms
You will often see Yfs as say 5000 uS. That's u = micro.
rds_on = 1/Yfs = 1/5000e-6 = 200 ohms
OH and mho is the same as Siemens; ohm spelt backwards!
Thanks Rob! I've never heard of Siemens, except for the brand. (Not mentioning Navy dudes or anything else.)
So maybe a "typical" 2SK30 would have an Rds-ON of around 500 ohms? Just ballpark guessing what the Yfs range might be, since they only spec a minimum.
QuoteSo maybe a "typical" 2SK30 would have an Rds-ON of around 500 ohms? Just ballpark guessing what the Yfs range might be, since they only spec a minimum.
I've got a rough number of rds_on = 440 ohms or Yfs = 2270uS.
The 2SK30A's are graded in 2SK30A-Y and 2SK30A-GR etc. If you dig though a number of datasheets you can piece more info together. You can also work stuff out from the graphs.
https://datasheetspdf.com/pdf/1392303/Toshiba/2SK30A/1
The Yfs stays relatively constant.
I'd say -Y would be Idss = 2mA VP = 1.8V, -GR would be Idss = 4mA, Vp = 3.5V.
I've got much better data for 2SK30A on a different computer, including measurements,
Only treat the above as a first-pass rough guide.
I've posted info on this forum in the past but it's often hard to find.
Re 2SK30A's, see this thread. Pretty sure I'm quoting from my larger pool of data,
https://www.diystompboxes.com/smfforum/index.php?topic=123153.0
reply #1:
"IIRC the 2SK30A-GR ends up with VP's around 3V typical."
reply #3:
"Whereas the 2SK30A-GR were more like 350 ohms or so. "
Those two values would imply Idss = 4.29mA for the 2SK30A-GR.
I've seen 2SK30A-O as Tube Screamer switches. Much lower Idss, but I don't know how that relates to Rds-ON. (It might be explained above, but I haven't wrapped my head around all of that yet. I've been busy repairing a speaker cable the last few hours, lol, had to re-stake the innards of a plug as it was a bit loose and adding about 25 ohms.)
QuoteI've seen 2SK30A-O as Tube Screamer switches.
They do pop-up. I know I've got some less common parts stashed away. I did take measurements at one point.
QuoteMuch lower Idss, but I don't know how that relates to Rds-ON. (It might be explained above, but I haven't wrapped my head around all of that yet.
Yes -O lower Idss. There is a table at the bottom of the first pages of the Japanese datasheet I linked. Typical Idss is probably 0.9mA for the -O.
As a ball-park figure Rds_on (ie. Yfs) doesn't change much across the categories. Rds_on has the least variation, then Vp and Idss has the worst variation. This comes the physics of the JFETs. That's why the categories sort by Idss. If you sort by the worst parameter the others automatically become constrained.
When you make a lot of measurements you can see some correlations between the parameters. You will pretty much never the a JFET with the smallest VP and the largest Idss. For one, the Yfs range limits the ratio of Idss and VP but even then you don't see extremes.
IIRC there's some graphs in this databook which shows (or lets you work out) the correllation between the parameters. It's one of the few places I've seen data like that written down,
https://archive.org/details/NationalSemiconductorFetDatabook1977
First of all, goes here below a list of the jFET's I have and the values (the closest to reaiity) that I measured of Vp and Idss (just important for comparing):
1: -1.304 V, 2.29 mA
2: -1.283 V, 2.36 mA
3: -1.306 V, 2.22 mA
4: -1.276 V, 2.33 mA
5: -1.289 V, 2.36 mA
6: -1.270 V, 2.28 mA
7: -1.302 V, 2.41 mA
8: -1.288 V, 2.36 mA
I understood that I had a big measurement error because of the limitations of my multimeter.
So, as for the Eb7+9 indication, I calculated the values of rdson with the equation rdson=(Vp)/(-2Idss), as follows:
1: 284 ohm
2: 271 ohm
3: 294 ohm
4: 273 ohm
5: 273 ohm
6: 278 ohm
7: 270 ohm
8: 272 ohm
And as PRR indicated, I measured voltages Vdrain of the jFET's, with a 9V battery connected to 100k ohm and following, the respective jFET. In this case, my battery indicated 9.31 V and the resistor was 98.0 kohm, giving me 95uA.
Follows the measurements:
• 1: 31 mV -> 326.31 ohm
• 2: 31.3 mV -> 329.47 ohm
• 3: 30.7 mV -> 323 ohm
• 4: 31.3 mV -> 329.47 ohm
• 5: 31.1 mV -> 327.37 ohm
• 6: 31.4 mV -> 330.53 ohm
• 7: 30.8 mV -> 324.2 ohm
• 8: 31.3 mV -> 329.47 ohm
I also connected a 1Megaohm to see the differences, and with a current of 9.28 uA, I got:
1 - 3.0 mV -> 323.27 ohm
2 - 3.0 mV -> 323.27 ohm
3 - 3.0 mV -> 323.27 ohm
So I concluded the difference would not be too big if I'd done that with the other FET's.
All of this seems very pretty, however with the first measurement I got around 325 ohm, but with the approximate calculations (taken of the Eb7+9 recommendated paper), I got around 275 ohm, so I wouldn't know which one of these would be the most accurate.. I'll assume that the measurements would be more realistic (given that I also did them with 1Megaohm and there wasn't a big difference).
What should I be trusting?
QuoteI understood that I had a big measurement error because of the limitations of my multimeter.
In what way?
When you measure stuff it's good assign an error to the measurement. From that you can work out how much error you expect in the estimates of rds_on.
If you have 2% error in VP and 10% error in IDSS then rds_on will have say 12% error.
Another issue is the meter affects the measurement. In this case you need to remove the effect of the meter. It's a bit like how it is possible to correct the VP measurements from RG JFET matcher to get a better VP value.
There's other factors as well:
- if you don't have high enough VDS IDSS can be low.
- if you have a high VDS then channel modulation can increase IDSS.
(more precisely it increases the measured ID in the IDSS test circuit.)
- when you measure IDSS the JFET can heat up.
https://www.vishay.com/docs/70596/70596.pdf
IDSS should decrease with increasing temperature
- the ohmic RS value will case an "internal" VGS voltage than will reduce IDSS.
Your estimates are showing lower rds_on based on VP and IDSS which means IDSS is low.
- When you use rdson=(Vp)/(-2Idss) it assumes a JFET mathematical model. How accurate is that?
QuoteWhat should I be trusting?
Rds_on is a direct measurement. It also includes the ohmic RD and RS values, see the text in my LTspice sim a few post back.
Here's a simulation estimating rds_on with different test set-ups:
- different *test circuit* RD values for VP; green = 1M, blue = 10M, red = 100M
- different VDS voltages for IDSS.
You can see that rds_on is low. That's caused by the channel modulation Lambda causing the ID measurement to be higher than the true IDSS.
It also shows if you use a 1M ohm impedance meter for RD the rds_on estimate is low because the VP estimate is low. This one you can apply a correction factor.
(https://i.postimg.cc/06XgtCz9/JFET-rds-on-from-IDSS-VP-2020-09-30.png) (https://postimg.cc/06XgtCz9)
Edit: FWIW, RD should be RS.
PRR is correct. All the others are way overkill. Good reading but not good answers. I have also noticed that all the others
are mixing apples and oranges. rds is not the same as Rd. The convention in electrical engineering and for that matter
the scientific world community in general is:
lower case letters are reserved for small signal (dynamic values).
UPPER CASE LETTERS ARE RESERVED FOR LARGE SIGNAL (STATIC VALUES).
1/gm is listed if you glean the DS. If your math is good from the DS you can derive all you need.
Also like one member has stated; two general cats.
Switching devices and amplification devices. What do you need?
If its amp then as stated, rds is irrelavent.
If its switch the DS will have it. Its that simple.
Thanks PRR.
Rob
Quoterds is not the same as Rd. The convention in electrical engineering and for that matter
the scientific world community in general is:
lower case letters are reserved for small signal (dynamic values).
UPPER CASE LETTERS ARE RESERVED FOR LARGE SIGNAL (STATIC VALUES)
RD (or RS) has not been confused with rds.
rds is the small signal resistance which follows the common JFET equations.
RD is an internal *ohmic* resistance. It's part of the SPICE model. You can see it on page 151,
https://www.seas.upenn.edu/~jan/spice/PSpice_ReferenceguideOrCAD.pdf
Quote from: Amptroll on September 30, 2020, 12:57:45 AM
What do you need?...If its switch the DS will have it. Its that simple.
Well not necessarily, if an amplifier FET has been used for switching, such as the 2SK30A in the TS9/808.
(And for the record, everyone, no I'm not going to go stick 330r resistors in my true-bypass TS clones. It's just something I've wondered about.)
Okk! Thank you very much Rob! My multimeter has a internal impedance of 10Mohm so that's in the middle (more like the 100Mohm),and it seems to me that the values I got are fair enough.. Today I'm gonna do real measurements with the oscilloscope on my university amd maybe I can confirm/check the values I got from my FET'S.
Another thing, when I measured the FET's, I always tried to do it as fast as I could, knowing that it could damage them. Does this help not to affect Idss as you mentioned? If not then maybe yes Idss was affected.
QuoteAnother thing, when I measured the FET's, I always tried to do it as fast as I could, knowing that it could damage them. Does this help not to affect Idss as you mentioned? If not then maybe yes Idss was affected.
It's a good idea.
Some switching JFETs have very high IDSS and low rds_on. You pretty much can't prevent the JFET from heating up. For these cases you can do a low current measurements (but not too low) and fit the JFET equation to extract IDSS. More than one measurement helps accuracy.
Manufacturers use pulse measurement techniques to prevent devices heating up but it requires more hardware to get accurate measurements.
Quote from: Rob Strand on September 30, 2020, 05:52:33 AM
It's a good idea.
Some switching JFETs have very high IDSS and low rds_on. You pretty much can't prevent the JFET from heating up. For these cases you can do a low current measurements (but not too low) and fit the JFET equation to extract IDSS. More than one measurement helps accuracy.
Manufacturers use pulse measurement techniques to prevent devices heating up but it requires more hardware to get accurate measurements.
[/quote]
Ok I understand. I will try to discover the jFET caractheristics by real circuit testing (based on the calculations I've done), maybe testing the frequency response with diferent Gate Voltages.