Hi,
I need to match two JFET for a compressor build. I use BF545A that is the SMD version of BF245A. And I use this circuit to plot the fet's:
(https://i.postimg.cc/gwXQgPXj/fetmatch.gif) (https://postimg.cc/gwXQgPXj)
At 0 VGS I can read more or less 0.24V VDS, but when I increment the voltage to 0.25 VGS, the result of VDS voltage is the same of the power supply... so... How I can plot this jfet? I'm doing something wrong?
I post the results of the mason audio. They use BF245A, and the voltage between 0 and 0.25 VGS are very different and more logical for me.
(https://i.postimg.cc/LYsTxt0T/mason.png) (https://postimg.cc/LYsTxt0T)
For a compressor or phaser that matching scheme might not be the best. A compressor or phaser operates the JFET near VDS = 0 upto say 1V max, in the resistance controlled region. The test jig produces a lot of the data points with VDS > 1V.
For accurate results ideally you need to use that jig with two multimeters as the DMM loads the gate supply. The loading isn't much. For a compressor you might get away with it but for a phaser you are pushing your luck a bit.
QuoteAt 0 VGS I can read more or less 0.24V VDS, but when I increment the voltage to 0.25 VGS, the result of VDS voltage is the same of the power supply... so... How I can plot this jfet? I'm doing something wrong?
When VGS is near zero the JFET is on and pulls down VDS. When VGS gets close or more negative than VP (Vgs off) the JFET will turn off and the VDS voltage will rise to just under the supply voltage. This is to be expected. VP is the point where VDS is near the supply. Your VP is around -1.75V to -1.5V, which looks OK for these JFETs.
At VGS near zero the voltage you measure is representative of the minimum resistance of the JFET. For example VD is around 0.25V, rds0 = Vds/ Id; Id = (9.58V-0.25V)/10k = 933uA. Vds = 0.25, rds0 = 268 ohm, which is very typical value for the minimum resistance.
Once you reach VP the datapoints aren't of much use. In fact once VDS reaches about 1V they aren't of much use because the JFET is operating a mode different to the final circuit. At VDS = 1V and Rd=10k the JFET resistance is only going to be around 1k ohm. Depending on the actual compressor circuit that might be OK, however, you might want to match the JFET resistance out to 3k ohm or 5k ohm. To match 3k ohm to 5k ohm and keep VGS < 1V you will need to increase the drain resistor to around 33k.
Instead of taking heaps of measurements you could take a few, for example,
- Vds @ Vgs=0, Rd = 10k ;rds0 = rds_min
- Vgs @ Vds=1V, Rd = 10k ;rds ~ 1.2k ohm
- Vgs @ Vds=1V, Rd = 33k ;rds ~ 4k ohm
This is my results.
(https://i.postimg.cc/xcnbjYLz/Screenshot-20230204-224213.png) (https://postimg.cc/xcnbjYLz)
This is enough for matching?
QuoteThis is enough for matching?
There's not enough information!
Firstly, the problem with your results is the JFETs have *much* lower VP than the common BF245A. Some of the JFETs are turning off at -0.25V and others at -0.5V. The BF245A's the VP values is around say -1.75V. The low VP JFET are common on ebay these days, they are more like switching JFETs. You can see the VDS voltage at VGS=0 is lower than the BF245A, that means the JFETs have a lower resistance to the BF245A's.
For a compressor a low VP means the compressor will distort for large signals. In fact it will start to distort with input voltages around the VP.
The low VP also means stepping at 0.25V doesn't give you many useful points for matching. The method I gave only finds useful points.
You have to ask yourself what is the point of matching?
More often than not you want the levels on a two-channel compressor to track and that means you want the outputs to be within 1dB or 0.5dB over the entire range of inputs.
The range of JFET resistances that need matching depends on the circuit!
If your circuit is asking for JFETs with VP's of -1.75V and you throw in JFETs with VP's around -0.5V then you already have a problem with the range of input voltages the circuit can handle. (and for some circuits the compression threshold is screwed up.)
The best of the bad might be Devices 8 and 20. High VP and match on two points, although only single digit resolution on VGS=0.
It may be easier to first test your JFETs for pinch-off voltage (Vgs(off)) and saturation current (Idss) to get potential pairs and then use a bridge type version of your test circuit to check how closely matched they are across the Vgs range. You can make and use quite simple circuits to do these tests using just a few basic components. If you are going to use a lot of JFETs in the future then it may be worthwhile investing in a Peak DCA75 semiconductor tester that will allow you to plot the characteristics curves on a PC.
For the time being, you may find the following circuits of some use for testing JFETs.
(https://i.postimg.cc/Z9fKfjMv/JFET-Test-and-Match-Circuits.png) (https://postimg.cc/Z9fKfjMv)
Honestly though, given the trouble getting hold of the damn things and the huge variability between supposedly-similar parts, why are we bothering with JFETs when we could perhaps be spending the effort researching better alternatives?
Seems to me that JFETs have had their day and we need to look elsewhere. I *know* there's lots of vintage circuits that use them, but they had those parts available in the 1970's and no other options, and now we *don't* have those parts available and we *do* have other options, so why are we trying to do the same thing they did way back then?!?
Your bf545's are all at the very low end of the Vgsoff specification. That is either bad luck, or could they be fake? You never know these days..
What is the schematic of your compressor? maybe your Jfets can be used with some changes to the bias arrangement.
Quote from: Clint Eastwood on February 05, 2023, 06:09:36 AM
Your bf545's are all at the very low end of the Vgsoff specification. That is either bad luck, or could they be fake? You never know these days..
What is the schematic of your compressor? maybe your Jfets can be used with some changes to the bias arrangement.
they are smd label 20, from nxp:
https://www.nxp.com/docs/en/data-sheet/BF545A_BF545B_BF545C.pdf
fake smd!?!? really?? It's for a 1176 compressor.
I've used this bf545a for a Dimension D clone also.
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus.html
But with this build, the jfet are for switching, so I think that it's not crucial the specs of the jfet here.
By coincidence, I have recently been looking at this...
https://www.masonaudio.org/diy/comp1176
Matching is only so the metering has a better chance of indicating what it's doing. Would it be so terrible if it didn't always?
Anyway, there's a matching technique described but unfortunately, the link for the schematic of the jig fails 404.
Quote from: spoontex on February 05, 2023, 06:41:43 AM
fake smd!?!? really??
Yes, there are loads of fake SMD parts including JFETs.
Quote from: anotherjim on February 05, 2023, 06:54:33 AM
By coincidence, I have recently been looking at this...
https://www.masonaudio.org/diy/comp1176
Matching is only so the metering has a better chance of indicating what it's doing. Would it be so terrible if it didn't always?
Anyway, there's a matching technique described but unfortunately, the link for the schematic of the jig fails 404.
The link is the shematic that I attach on the first post.
Hmm, looking at the schematic I don't see an easy way to use low Vgsoff jfets like you have.
The j113 has a Vgsoff of around -1.6 volts, and in my experience they are pretty consistent. And she has legs! The transconductance is a bit higher than a bf245a, but I don't see why it would be a problem.
Quote from: Clint Eastwood on February 05, 2023, 08:23:43 AM
Hmm, looking at the schematic I don't see an easy way to use low Vgsoff jfets like you have.
The j113 has a Vgsoff of around -1.6 volts, and in my experience they are pretty consistent. And she has legs! The transconductance is a bit higher than a bf245a, but I don't see why it would be a problem.
what scheme are you referring to?
So, I need to rechange all my fet's on my Dim D build. What jfet can I use? J113? Because found some NOS BF245A, but they are expensive... The original uses 2sk30, that can I found at reasonable price. If not, I can use 2n5457 smd version, for both, Dim D and 1176.
And I can use this "fake" BF545A as input buffer?
The whole thing (possibly?)...
(http://www.gyraf.dk/gy_pd/1176/1176sch.gif)
Quote from: ElectricDruid on February 04, 2023, 07:27:55 PM
Seems to me that JFETs have had their day and we need to look elsewhere.
You could say that about tubes, germanium transistors and vinyl records. Someone always wants that 'less than perfect' sound.
Quote from: spoontex on February 05, 2023, 10:33:40 AM
what scheme are you referring to?
So, I need to rechange all my fet's on my Dim D build. What jfet can I use? J113? Because found some NOS BF245A, but they are expensive... The original uses 2sk30, that can I found at reasonable price. If not, I can use 2n5457 smd version, for both, Dim D and 1176.
And I can use this "fake" BF545A as input buffer?
I was referring to the schematic posted above by anotherjim. The J113 has higher Idss and transconductance than 2sk30 and BF245a. The 2n5457 would be a better replacement I think.
You can use your BF545A's as buffer, but you cannot use the auto-bias arrangement with just a source resistor. Your Vgsoff is only 0.5V, if your signal would go above that it will be clipped at the bottom. So you should bias the gate at least at +1volt or so with a voltage divider.
Quote from: Clint Eastwood on February 05, 2023, 01:25:14 PM
Quote from: spoontex on February 05, 2023, 10:33:40 AM
what scheme are you referring to?
So, I need to rechange all my fet's on my Dim D build. What jfet can I use? J113? Because found some NOS BF245A, but they are expensive... The original uses 2sk30, that can I found at reasonable price. If not, I can use 2n5457 smd version, for both, Dim D and 1176.
And I can use this "fake" BF545A as input buffer?
I was referring to the schematic posted above by anotherjim. The J113 has higher Idss and transconductance than 2sk30 and BF245a. The 2n5457 would be a better replacement I think.
You can use your BF545A's as buffer, but you cannot use the auto-bias arrangement with just a source resistor. Your Vgsoff is only 0.5V, if your signal would go above that it will be clipped at the bottom. So you should bias the gate at least at +1volt or so with a voltage divider.
And for the Dim D? I can leave the BF545A? or I need to change?
Do you have a schematic of the Dim D?
Quote from: Clint Eastwood on February 05, 2023, 01:57:23 PM
Do you have a schematic of the Dim D?
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus.html
Main board:
PSU and LFO (HA1475 version not shown separately)
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg1.pdf
Input Section (HA1475 version not shown separately)
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg2.pdf
Output Section (HA1457 + Transistors version)
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg3a.pdf
BBD Board:
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus_BBD_Board_sch_page1.pdf
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus_BBD_Board_sch_page2.pdf
So you have built this, right? then the question is simply : does it work?
To be honest, I don't feel confident to tell you if it would work with your BF545's just by looking at the schematic.
Quote from: Clint Eastwood on February 05, 2023, 02:23:16 PM
So you have built this, right? then the question is simply : does it work?
To be honest, I don't feel confident to tell you if it would work with your BF545's just by looking at the schematic.
It work! I use every day. It's strange because the bag of this bf545a says clearly NXP, it don't seems fake. I think that only have bad specs, I don't know.
Quote from: spoontex on February 05, 2023, 02:43:36 PM
It's strange because the bag of this bf545a says clearly NXP, it don't seems fake.
I have had some experience with what were meant to be NXP BF862 JFETs. The datasheet shows that the marking code is '2Ap' however it seems that there are also ones marked as '2AW'.
Both markings are correct for the NXP BF862. It seems that the dies were made in a European NXP fab but they were encapsulated in two places, Hong Kong for the '2Ap' and China for the '2AW'.
Anyway, I bought both versions.
The datasheet shows that Vgs(off) has a typical value of -0.8v.
On testing I found that all of the '2Ap' ones measured about -3.6v to -3.8v.
Half of the '2AW' ones measured about -3.8v as well however there were some '2AW' ones that measured around -0.7v to -0.75v which would agree with the datasheet.
So there must be some fake NXP devices around as those values are way off even for out of spec devices.
QuoteIt's strange because the bag of this bf545a says clearly NXP, it don't seems fake.
What's the markings on the devices:
According to the datasheet they should be 20p, or 20t, or 20W.
From your test results the devices have:
- low resistance (100 ohm)
- low VP
These aren't characteristics of BF545A's. They are characteristics of the generic stuff you get off ebay.
Quote from: Rob Strand on February 05, 2023, 04:01:00 PM
QuoteIt's strange because the bag of this bf545a says clearly NXP, it don't seems fake.
What's the markings on the devices:
According to the datasheet they should be 20p, or 20t, or 20W.
From your test results the devices have:
- low resistance (100 ohm)
- low VP
These aren't characteristics of BF545A's. They are characteristics of the generic stuff you get off ebay.
(https://i.postimg.cc/xXVtYCPS/2023-2-5-22-19-12.jpg) (https://postimg.cc/xXVtYCPS)
(https://i.postimg.cc/QVBbT543/2023-2-5-22-19-55.jpg) (https://postimg.cc/QVBbT543)
What modern replacement can I use??
QuoteWhat modern replacement can I use??
They look like the real thing.
You might try testing the VP values with a different test jig. At least that gives you a second opinion.
https://runoffgroove.com/fetzervalve.html
The problem isn't what you buy who you are buying off.
There are legitimate suppliers out there who are trying to shut down fakes and re-marked components. You should be able contact your supplier, find the guy that handles fakes, and explain the issue. The legit supplies want to weed out their bad suppliers as well. It's likely they will organize a swap.
For the non-legit suppliers, it's hit and miss. You might buy another part number and end-up with the exact same problem. There's many posts on this forum where people have build problems with low VP JFETs. It's a widespread problem now for JFETs.
Onsemi makes the MMBFj309LT1G, wich is an smd version of the J309, wich is equivalent to the BF245a. Mouser Europe has 450.000 of them in stock (!)
Edit: Actually, it is equivalent to the BF245b.
Quote from: ElectricDruid on February 04, 2023, 07:27:55 PM
Honestly though, given the trouble getting hold of the damn things and the huge variability between supposedly-similar parts, why are we bothering with JFETs when we could perhaps be spending the effort researching better alternatives?
Seems to me that JFETs have had their day and we need to look elsewhere. I *know* there's lots of vintage circuits that use them, but they had those parts available in the 1970's and no other options, and now we *don't* have those parts available and we *do* have other options, so why are we trying to do the same thing they did way back then?!?
I completely agree. there are many decent and simple alternatives. VCA and a compressor that works better and doesn't clutter the sound. OTA in feedback mode, like in Engineers thumb. Optical compressors for coloration and natural sound.
Quote from: FSFX on February 05, 2023, 12:37:36 PM
Quote from: ElectricDruid on February 04, 2023, 07:27:55 PM
Seems to me that JFETs have had their day and we need to look elsewhere.
You could say that about tubes, germanium transistors and vinyl records. Someone always wants that 'less than perfect' sound.
It makes sense. look at how the lamp teacher industry is changing. more and more manufacturers are making compact 20W amplifiers, but with tubes from 50W amplifiers. more and more concentrating musicians take Kemper, Ax Fx, Neural DSP to concerts. germanium transistors, they are becoming less and less common. in my country they were produced until the end of the 90s and they are very cheap, they are still sold in stores (not for guitar diy, but simple radio parts stores). but when they run out there will be new transistors (I bet on MOSFet).
vinyl is popular again. but I'm too used to music services and the ability to instantly change songs and artists. I have a childhood friend who owns a vintage equipment repair business from Eastern Europe. he says that warm and good sound is not vinyl, vinyl is bad, compressed and modern, haha good sound is the bobbin.
I know, modern times... streaming, kemper and that all stuff that is amazing. But the fact is that I have a device with 14 supicious fet's that it need to change. I've two alternatives for the Dim D build
2sk30, at 0,40€/u
2n5957, smd with adapter 1€/u
For the 1176 build, I buy a matched pair. It will be more economic.
What do you think It's better replacement bf245a for the dim d?
Toshiba still make some jfets, the 2sk208 looks like a good , probably even better than the original replacement for BF245a. Look up the datasheet.
Quote from: Clint Eastwood on February 06, 2023, 03:54:03 AM
Toshiba still make some jfets . . .
There are companies like InterFET that specialise in making JFETs. It is their core business as their name implies.
History has made the Urie 1176 a very desirable device. If it isn't JFET, it isn't an 1176.
If the JFETs are socketed, you can sort out matching anytime. Put the best one for the compression in for keeps and wait to get a better match for the metering. I cannot understand the concern about the metering. The metering JFET is in a different circuit - it's a source follower and there are two trimmers to set up the response.
Quote from: FSFX on February 06, 2023, 05:10:26 AM
Quote from: Clint Eastwood on February 06, 2023, 03:54:03 AM
Toshiba still make some jfets . . .
There are companies like InterFET that specialise in making JFETs. It is their core business as their name implies.
Yes, but their fets cost 10 times more..
Quote from: Clint Eastwood on February 06, 2023, 07:13:21 AM
Yes, but their fets cost 10 times more..
Just like NOS germanium transistors then?
Quote from: anotherjim on February 06, 2023, 06:21:34 AM
History has made the Urie 1176 a very desirable device. If it isn't JFET, it isn't an 1176.
If the JFETs are socketed, you can sort out matching anytime. Put the best one for the compression in for keeps and wait to get a better match for the metering. I cannot understand the concern about the metering. The metering JFET is in a different circuit - it's a source follower and there are two trimmers to set up the response.
I need to match for a stereo 1176.
There is always the PWM phaser that depends on duty-cycle modulating analog switches to mimic JFET's operated in the variable resistance mode:
(https://i.postimg.cc/y6cBvJ0n/PWMEnv-Phaser-V1-3-1.png)
It uses a triangle wave generator operating above audio frequencies to set the switching on-off point of analog switches. This is modulated by the LFO to lengthen or shorten the the ON time. The advantage? The JFET's are replaced by CD4066 devices and there is no need to select devices. The disadvantage is the high-frequency oscillator means that if you want to operate from AC power, you have to pass electromagnetic compatibility testing to be able to sell it.
Quote from: Clint Eastwood on February 06, 2023, 07:13:21 AM
Yes, but their fets cost 10 times more..
If cost is the only concern then if the OP wants to messages me with their address I will post off a strip of 20 genuine ON Semi MMBFJ113 JFETs (marking code H_59) to them for free.
A sample of them I tested showed Vgs(off) = -1.06v. Idss = 8.56mA and Ron = 49 ohms.
There, despite all the flack I keep getting here, I can't be fairer than that.
EDIT: I also have TO-92 THT versions if they are preferable to using the SMD ones.
Quote from: spoontex on February 06, 2023, 07:24:00 AM
Quote from: anotherjim on February 06, 2023, 06:21:34 AM
History has made the Urie 1176 a very desirable device. If it isn't JFET, it isn't an 1176.
If the JFETs are socketed, you can sort out matching anytime. Put the best one for the compression in for keeps and wait to get a better match for the metering. I cannot understand the concern about the metering. The metering JFET is in a different circuit - it's a source follower and there are two trimmers to set up the response.
I need to match for a stereo 1176.
Ah, Stereo! I'd find making just one channel a major project. Although if you have PCB already worked out, it isn't so bad.
Have to say I don't understand JFET types. Some are marketed as switches and some as signal amplifiers. Are any specifically made to be voltage-controlled resistors? I would think the amplifier types were closest but I think the 2SK30A has been used in all three roles.
Quote from: anotherjim on February 06, 2023, 08:13:01 AM
Have to say I don't understand JFET types. Some are marketed as switches and some as signal amplifiers.
Have you read any of these ON Semi documents?
https://www.onsemi.jp/download/application-notes/pdf/an-6609.pdf
Quote from: anotherjim on February 06, 2023, 08:13:01 AM
Are any specifically made to be voltage-controlled resistors?
The answer is yes. Check out Linear Systems - here is an extract from a document of theirs -
Linear Systems has expanded its family of JFET Voltage-Controlled Resistors (VCRs) for reduced-power circuits over IC-based devices.
Two new additions to the JFET Voltage-Controlled Resistor Family include the LS26VNS, a Single N-Channel JFET VCR, and the LS26VPS, a Single P-Channel JFET VCR. Both the N and P-channel versions are made from the same die geometry, making them complimentary.
They are in addition to the existing VCR11N, Dual N-Channel.
The advantages of using VCRs for voltage-controlled amplifier or voltage-controlled gain systems include:
- VCR-based devices operate with lower voltage and power consumption, enabling battery-powered systems to last significantly longer.
- VCR JFETs can be used with low power op amps to provide a low power solution for voltage-controlled gain systems.
- With low voltage op amps such as the LMV358 or TLC272, VCR JFETs are used to perform voltage-controlled gain at extremely low supply voltages such as +/- 2 volts, + 5 volts and even down to a single 3.3 volt power supply.
- Circuits can be designed for very low power consumption when low voltage, low power op amps are included with the JFET VCRs. One example would be to include using an LSK389A as a VCR for these low voltage circuits.
- The matched pair JFET VCRs (VCR11N or LSK389A) can provide two identical channels of voltage-controlled gain where the control voltage tracks for both channels. This can be used in a stereo audio signal attenuator.
- Typical active voltage-controlled amplifiers drain more current and require higher minimum power supply voltages needing at least +/- 4 volts for a specialized voltage control amplifier integrated circuit.
- P-Channel JFETs including a J177 can be designed as a VCR circuit with a single a power supply. For example, the P-Channel JFET VCR circuit is ideal with positive supply voltages since it's control voltage to the gate is also a positive voltage.
There is a Vishay application note regarding the use of FETs as VCRs.
https://www.vishay.com/docs/70598/70598.pdf
Please, anybody can help me to select a replacement for the bf245a?
witch one?
2sk30a
j113
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus.html
Main Board:
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg1.pdf
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg2.pdf
http://jhaible.com/legacy/subtle_chorus/subtle_chorus_main_board_sch_dwg3a.pdf
BBD Board:
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus_BBD_Board_sch_page1.pdf
http://jhaible.com/legacy/subtle_chorus/jh_subtle_chorus_BBD_Board_sch_page2.pdf
Thanks.
From designer old forum:
"I haven't tried a BF245C, but I specified a BF245A for a reason: The threshold voltages are vastly different."
If you look up the datasheets, you can choose yourself. Look for Vgsoff similar to BF245a. Both 2SK30 and J113 should work I think. FSFX made you a kind offer, I would bite if I were you ;)
At last I found good BF245A!
Here are the results. I need to match two of them. What would be good candidates? This 10 devices are enough? I have more of them...
(https://i.postimg.cc/5X24wr9B/results.png) (https://postimg.cc/5X24wr9B)
(https://i.postimg.cc/KKgyfS6M/grafic.png) (https://postimg.cc/KKgyfS6M)
Quote from: spoontex on March 01, 2023, 04:54:20 AM
What would be good candidates? This 10 devices are enough? I have more of them...
It looks like devices 4, 5 and 6 are the closest matches.
BTW, not sure how you tested but you seemed to be restricting the current to less than 1mA so this has affected the accuracy of your data for higher current. As the Idss for a 2N245A is going to be between 2mA and 6.5 mA then you are artificially restricting the current so the data at the top of your graph is invalid.
(https://i.postimg.cc/Y4CFsyH0/grafic2.png) (https://postimg.cc/Y4CFsyH0)
Quote from: FSFX on March 01, 2023, 07:43:46 AM
Quote from: spoontex on March 01, 2023, 04:54:20 AM
What would be good candidates? This 10 devices are enough? I have more of them...
It looks like devices 4, 5 and 6 are the closest matches.
BTW, not sure how you tested but you seemed to be restricting the current to less than 1mA so this has affected the accuracy of your data for higher current. As the Idss for a 2N245A is going to be between 2mA and 6.5 mA then you are artificially restricting the current so the data at the top of your graph is invalid.
(https://i.postimg.cc/Y4CFsyH0/grafic2.png) (https://postimg.cc/Y4CFsyH0)
with this circuit:
https://www.axtsystems.com/joomla31/index.php/projects/11-cat-1176ln-audio-compressor/4-art-1176-fet-matching