Looking to modify a transistor input stage (LTP), and thought that adding a current source using a CRD would be an interesting thing to experiment with.
Seems simple...remove the common emitter resistor, replace with a two-lead CRD rated for the desired current.
Besides the PITA to find these devices, I can see no real reason why they are not used more often. A current source for a LTP is so much superior to a resistor for biasing, and the easiest way to go about it is using a CRD. So I can't help but wonder why I do not see these devices used as I have described....?
Anyone try designing using CRD's? Problems?
Lastly, has anyone got a source for the Vishay CRD's? Excellent specs and very low output capacitance (6pf or so), but no one seems to carry them....
Seems simple...remove the common emitter resistor, replace with a two-lead CRD rated for the desired current.
Besides the PITA to find these devices, I can see no real reason why they are not used more often. A current source for a LTP is so much superior to a resistor for biasing, and the easiest way to go about it is using a CRD. So I can't help but wonder why I do not see these devices used as I have described....?
Anyone try designing using CRD's? Problems?
Lastly, has anyone got a source for the Vishay CRD's? Excellent specs and very low output capacitance (6pf or so), but no one seems to carry them....
EchoWars said:
Are you talking about constant current diodes?
If you are maybe they are not too much used because they are expensive, not easy to find and not as precise as they should be.
A FET + adjusting resistor, preset and hand picked for exact current, might be a better alternative to what you proppose. For the same objective. Maybe.
Carlos
J505
Hi ,
You mean J505 etc. thingies?
Look here:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=34639&highlight=J505&r=&session=
John Curl says in reply to my post these are noisy and useless except for high level circuits:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=28040&highlight=curl+current+source&r=&session=
You can also use a LED and a transistor. Less noisy!
The link should be to www.analog.com for the MAT03 (LED current source) and MAT02 (two transistor current source) datasheet. [All typo's in my old post I later discovered]
Hi ,
You mean J505 etc. thingies?
Look here:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=34639&highlight=J505&r=&session=
John Curl says in reply to my post these are noisy and useless except for high level circuits:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=28040&highlight=curl+current+source&r=&session=
You can also use a LED and a transistor. Less noisy!
The link should be to www.analog.com for the MAT03 (LED current source) and MAT02 (two transistor current source) datasheet. [All typo's in my old post I later discovered]
Constant current diodes
Using JFETs + 1 resistor is much more flexible.
Those diodes are in fact JFETs, but you can not change the current,
as only the output pins are available.
-----------------------------------------------------------------
www.elfa.se has some current doides for different mA.
But in order to work they need >= 10 Volt across them.
That will make them unsuitable for some applications.
The diodes, E101 ...., also have wide tolerance +-20% from nominal value of milliampere.
example:
E202 mA= 1,68-2,32; Vf= 10-100Volt; Rdiff= 0,25MOhm
E103 mA= 8,00-12,0; Vf= 10-100Volt; Rdiff= 0,17MOhm
ELFA - constant curr diodes
------------------------------------------------------------------
Buy yourself 10-20 pieces of low noise JFETs.
And make your own 2-pin diodes out of them.
Much better!
/halo - the constant and current poster
Using JFETs + 1 resistor is much more flexible.
Those diodes are in fact JFETs, but you can not change the current,
as only the output pins are available.
-----------------------------------------------------------------
www.elfa.se has some current doides for different mA.
But in order to work they need >= 10 Volt across them.
That will make them unsuitable for some applications.
The diodes, E101 ...., also have wide tolerance +-20% from nominal value of milliampere.
example:
E202 mA= 1,68-2,32; Vf= 10-100Volt; Rdiff= 0,25MOhm
E103 mA= 8,00-12,0; Vf= 10-100Volt; Rdiff= 0,17MOhm
ELFA - constant curr diodes
------------------------------------------------------------------
Buy yourself 10-20 pieces of low noise JFETs.
And make your own 2-pin diodes out of them.
Much better!
/halo - the constant and current poster
Just like zener diodes, CRD's can be extremely convenient tools, and if you know how to deal with them, they can be useful for low-level as well as high-level circuits. However, they have certain quirks that should be taken into account. Careful studying of the catalogs (including the curves) should give you some idea of how these devices should be used (and also how they should _not_ be used).
In addition to Vishay-Siliconix, CRDs are made by Semitec-Ishizuka:
http://www.semitec.co.jp/pdf/crd/crdALL.pdf
Warning - the above pdf is in Japanese.
regards, jonathan carr
In addition to Vishay-Siliconix, CRDs are made by Semitec-Ishizuka:
http://www.semitec.co.jp/pdf/crd/crdALL.pdf
Warning - the above pdf is in Japanese.
regards, jonathan carr
http://www.vishay.com/document/70196/70196.pdf
Warning, the pdf above is in English.
It is simple. Why not use the tool designed for the job? You get excellent temperature stability, low capacitance - typically just a couple of pF, operation from 2 or 3V to 50V and fairly high Z (>200kohms).
Warning, the pdf above is in English.
It is simple. Why not use the tool designed for the job? You get excellent temperature stability, low capacitance - typically just a couple of pF, operation from 2 or 3V to 50V and fairly high Z (>200kohms).
Jonathan,
I swear I have to learn Japanese soon>
Another source of current diodes is Central Semiconductor. I used to purchase from them years ago.
http://www.centralsemi.com/
Jam
I swear I have to learn Japanese soon>
Another source of current diodes is Central Semiconductor. I used to purchase from them years ago.
http://www.centralsemi.com/
Jam
Well, finding a manufacturer is a hell of a lot easier than finding someone to sell you a few. I found a lot of devices online at manufacturer sites, but pretty damn few sellers.
The Vishay parts (CR160 et al) look nice, but I have yet to find a seller. Pretty tight tolerances on these.
If I decided to 'roll my own', might it be better to skip the JFET idea, and simply use a bipolar? I am still attracted by the two-pin CR160 series, but I hate to sacrifice noise parameters...
The Vishay parts (CR160 et al) look nice, but I have yet to find a seller. Pretty tight tolerances on these.
If I decided to 'roll my own', might it be better to skip the JFET idea, and simply use a bipolar? I am still attracted by the two-pin CR160 series, but I hate to sacrifice noise parameters...
EchoWars said:
It is nice to have 2 pin device.
And the JFET or CC-diode has one other obvious advantage.
It does not have to use a separate power supply.
It is only supplied by the current, that runs through it.
A bipolar (or a MOS) needs a reference voltage.
This can be made out of a red LED, another bipolar used
as current limiter, or some other reference.
This reference needs a supply.
This is usually taken via resistor from Power Supply.
And to avoid noise/hum from the supply
you can need to use filter caps Elyts and/or Film caps.
The need is depending on how clean the supply is.
But as the LED and bipolar smallsignal has low noise
and good temp-matching,
this can give very good, clean and very constant current.
-----------------------------------------------------------
So diode/JFET method is simple and does not need a ref with PSU.
It makes it isolated from Power Supply Rail disturbance.
LED + bipolar, or a bipolar with another bipolar as current limiter,
gives very clean low noise.
But they need contact with PSU-rails, and can require
heavy or less filtering, to decouple from PSU.
Apart from that you will never find any problems
to get a hold of LEDs or small-signal Bipolar Transistors
/halo - uses bipolar Constant Current Sources
(because he has not many JETs around, and not CCS-diodes)
I agree. But it does little good to know they exist when finding someone to sell them to you is nigh-on impossible. I can devise a bipolar CC source with three connection points on a daughterboard...nowhere near as neat and clean as the CRD solution, but what's a poor boy to do?traderbam said:
Hugh: My personal experience is that a good CCS both works better and sounds better than a resistor. Compared to a resistor, the current really can be made totally flat no matter what the load circuit is doing, the output impedance can be far higher (enough so that board parasitics are the limiting factor, rather than the circuit), and the output capacitances can be dealt with, too.
What would you give as the reasons for your preferring resistors?
regards, jonathan carr
What would you give as the reasons for your preferring resistors?
regards, jonathan carr
In the U.S.A. mouser has these http://www.mouser.com//index.cfm?ha...&criteria=1n5309&searchby=PartNumber&x=8&y=10
The Central Semi parts are rather unremarkable...sloppy tolerances and high output capacitance. They might work well as a test, but I'd really, really like to find some of the Vishay/Siliconix CRD's, specifically the CR160. Nice tolerances and way low C_out.
Avnet carries them. Min order of 200.
Avnet carries them. Min order of 200.
There is a Japanese firm called Wakamatsu that does mail-order sales of various electronic components. I haven't confirmed if they sell overseas, however. Go to:
http://www.wakamatsu.biz/
The link for the CRDs is:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=1206&page=0
They also sell Black Gates:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2103&page=0
Oscons:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2104&page=0
Cerafines:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2107&page=0
Silmics:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2108&page=0
Soshin SE mica caps (recommended by Kaneda):
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2101&page=0
Riken resistors 0.5W:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2005&page=0
Riken resistors 0.5W gold-plated leads:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2013&page=0
Riken resistors 2W:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2014&page=0
Skelton resistors (recommended by Kaneda):
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2008&page=0
And lots of other audiophile-approved favorites.
Wakamatsu even has the "Bowlingual", which supposedly interprets what your dog is trying to say. Japanese version only (maybe it is easier to learn Japanese than dog-talk?):
http://www.wakamatsu-net.com/cgibin/biz/psearch.cgi?page=0&word=ƒoƒEƒŠƒ“ƒKƒ‹
regards, jonathan carr
http://www.wakamatsu.biz/
The link for the CRDs is:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=1206&page=0
They also sell Black Gates:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2103&page=0
Oscons:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2104&page=0
Cerafines:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2107&page=0
Silmics:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2108&page=0
Soshin SE mica caps (recommended by Kaneda):
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2101&page=0
Riken resistors 0.5W:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2005&page=0
Riken resistors 0.5W gold-plated leads:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2013&page=0
Riken resistors 2W:
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2014&page=0
Skelton resistors (recommended by Kaneda):
http://www.wakamatsu-net.com/cgibin/biz/page.cgi?cate=2008&page=0
And lots of other audiophile-approved favorites.
Wakamatsu even has the "Bowlingual", which supposedly interprets what your dog is trying to say. Japanese version only (maybe it is easier to learn Japanese than dog-talk?):
http://www.wakamatsu-net.com/cgibin/biz/psearch.cgi?page=0&word=ƒoƒEƒŠƒ“ƒKƒ‹
regards, jonathan carr
An easy way to deal with the capacitence issue is a small series resistor between the CRD and the Load. this is also true with devices offering low capacitence, because the output C of any solid state device and most notably fet devices is non linear and will varry with changing signal and if your diff amp is used in a non inverting Amp there will be a signal at the emitter node of the Diff Amp that will change as the input signal changes.
As mentioned above a far better solution is to obtain a low noise low Cap fet transistor and connect the gate and source together or place a resistor between the gate and source to adjust the current to your liking. An evean better solution is to use two jfets in cascode or a BJT + jfet in cascode. however this i am sure more complex than you wish. so if you can not find the Brand of CRD you are looking for then the jfet transistor looks more attractive.
As mentioned above a far better solution is to obtain a low noise low Cap fet transistor and connect the gate and source together or place a resistor between the gate and source to adjust the current to your liking. An evean better solution is to use two jfets in cascode or a BJT + jfet in cascode. however this i am sure more complex than you wish. so if you can not find the Brand of CRD you are looking for then the jfet transistor looks more attractive.
Jonathan,
I've done comparisons, and so has Robert Jones in CA, and we both believe the sonics are superior; fatter, warmer. There is no perceived loss of detail, and distortion on such an amp is not measureably affected. Robert measured 0.0045% at 1KHz 8R 10W on one of my AKSAs.
This clearly is second harmonic distortion, caused by the asymmetric delivery of current from a simple resistor. This approach is only viable on a single diff input stage, since a fully complementary beast induces identical distortion profiles at both ends of the waveform, thus qualifying as odd order, and nasty.
Like much of my research, it is based on sonics, however, rather than measurement, and while I regret not having an AP 1 for painstaking analysis, I will gleefully admit to designing solely for sound quality. I'm actually surprised you've had the opposite experience, but hey, part of life's rich tapestry........!
Cheers,
Hugh
I've done comparisons, and so has Robert Jones in CA, and we both believe the sonics are superior; fatter, warmer. There is no perceived loss of detail, and distortion on such an amp is not measureably affected. Robert measured 0.0045% at 1KHz 8R 10W on one of my AKSAs.
This clearly is second harmonic distortion, caused by the asymmetric delivery of current from a simple resistor. This approach is only viable on a single diff input stage, since a fully complementary beast induces identical distortion profiles at both ends of the waveform, thus qualifying as odd order, and nasty.
Like much of my research, it is based on sonics, however, rather than measurement, and while I regret not having an AP 1 for painstaking analysis, I will gleefully admit to designing solely for sound quality. I'm actually surprised you've had the opposite experience, but hey, part of life's rich tapestry........!
Cheers,
Hugh
Hugh,
Using your same amplifier as a test bed have you tried doubleing the current through the differential wiith corresponding reduction of the collector resistor/s. You might find that that when you performed the experiment that the current source was set at too low a value.
I have to agree with Jonathan that in all the circuits that I have built the current source always won out . Maybe this is due to current starvation of the next stage? A resistor probably has more latitude on extreme swings than a current source set at the same value during static conditions.
Regards,
Jam
Using your same amplifier as a test bed have you tried doubleing the current through the differential wiith corresponding reduction of the collector resistor/s. You might find that that when you performed the experiment that the current source was set at too low a value.
I have to agree with Jonathan that in all the circuits that I have built the current source always won out . Maybe this is due to current starvation of the next stage? A resistor probably has more latitude on extreme swings than a current source set at the same value during static conditions.
Regards,
Jam
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.