mikep
Well-known member
I never heard about this till now. Do you think you could make a opto compressor with one of these? they are unbelievebly fast compared to a vactrol.
LS627 light-sensitive JFET
- Thread starter mikep
- Start date
Help Support GroupDIY Audio Forum:
It's a big Web. Give us a link.
http://www.linearsystems.com/datasheets/LS627.pdf
> I never heard about this till now. Do you think you could make a opto compressor with one of these?
Same as any FET limiter, only not as good.
The FET is a linear resistor only for zero voltage across the channel. But in a limiter, the voltage across the channel is our output; it can't be zero, not even real low.
In practice, we can go 5mV-500mV with "tolerable THD", depending on the device (especially Vth) and what we can tolerate.
A cute trick is to cancel some of the THD by injecting half the signal voltage at the gate (assuming a symmetrical FET, which is the norm). The implementation is semi-trivial. This is not trivially possible with the light control.....
Or is it? LS does pin-out the Gate. The requirement is to mirror the voltage on the channel at the average Gate position. So the THD-null signal does not have to be light and does not have to be proportional to light.
Maybe one of the silicon gurus will have a clue.
But when all is said and done: what have you gained? What advantage is there in light-isolation? There is an obvious disadvantage: the FET Gate is many-Megs and microWatts of drive power; the light drive is low-Z and milliWatts of power.
Speed is far faster than the chemical process in a photoresistor, but comparable to any JFET.
The real effect of the few-mV of signal possible at low THD is to limit your signal to noise ratio. Given ideal amplifiers, the idle noise is related to the size of series resistor before the FET, which in turn is proportional to the size of the FET. The maximum level is related to Vth, which is not a strong function of the FET size but is commonly minimized for good Amplifier and easy Switch use.
Odds are that you can find a hi-Vth lo-R non-opto JFET and get better audio performance than LS's one opto-FET.
BTW: this isn't new. One of the Japanese houses has been putting an LED and an FET in one package for a couple decades. I tried like heck to use it for a limiter, but the effective Vth was just too low.
http://www.linearsystems.com/datasheets/LS627.pdf
> I never heard about this till now. Do you think you could make a opto compressor with one of these?
Same as any FET limiter, only not as good.
The FET is a linear resistor only for zero voltage across the channel. But in a limiter, the voltage across the channel is our output; it can't be zero, not even real low.
In practice, we can go 5mV-500mV with "tolerable THD", depending on the device (especially Vth) and what we can tolerate.
A cute trick is to cancel some of the THD by injecting half the signal voltage at the gate (assuming a symmetrical FET, which is the norm). The implementation is semi-trivial. This is not trivially possible with the light control.....
Or is it? LS does pin-out the Gate. The requirement is to mirror the voltage on the channel at the average Gate position. So the THD-null signal does not have to be light and does not have to be proportional to light.
Maybe one of the silicon gurus will have a clue.
But when all is said and done: what have you gained? What advantage is there in light-isolation? There is an obvious disadvantage: the FET Gate is many-Megs and microWatts of drive power; the light drive is low-Z and milliWatts of power.
Speed is far faster than the chemical process in a photoresistor, but comparable to any JFET.
The real effect of the few-mV of signal possible at low THD is to limit your signal to noise ratio. Given ideal amplifiers, the idle noise is related to the size of series resistor before the FET, which in turn is proportional to the size of the FET. The maximum level is related to Vth, which is not a strong function of the FET size but is commonly minimized for good Amplifier and easy Switch use.
Odds are that you can find a hi-Vth lo-R non-opto JFET and get better audio performance than LS's one opto-FET.
BTW: this isn't new. One of the Japanese houses has been putting an LED and an FET in one package for a couple decades. I tried like heck to use it for a limiter, but the effective Vth was just too low.
H11F1 - H11F3 are old-news optocouplers based on this technology:
http://www.fairchildsemi.com/pf/H1/H11F3.html
Nice for remote switching of audio - I haven't found it useable for compression.
Jakob E.
http://www.fairchildsemi.com/pf/H1/H11F3.html
Nice for remote switching of audio - I haven't found it useable for compression.
Jakob E.
bcarso
Well-known member
It's not clear to me what the H11 series is doing to stay off in the absence of light. Perhaps someone knows what the internal structure really is. Maybe they are enhancement mode DMOS parts with photodiodes connected to the gates.
As far as the LS627 though---the benefits of photocontrol seem to be analogous to the real estate adage: isolation, isolation, and isolation. As PRR indicates, you can get rid of most of the second harmonic distortion with the feedback of half of drain voltage to the gate---but then what's the advantage of the isolation? If it's really just the single JFET shown, in order to keep the device off you need some connection to the gate to apply a negative bias. This has to be a big resistor or you kill your photosensitivity. But if large you will get distortion from the modulation of the channel-gate capacitance (coincidentally, I just saw this effect while making some improvements to a client's DMOS-FET based limiter---they were using a 2N7000, which to my surprise did fairly well in the app).
Where the photogeneration of drain-gate leakage shines (sorry couldn't resist) is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor. Landis et al. did this at Lawrence Berkeley in the 70's iirc. You can have as your only feedback element a tiny capacitor, and periodically pulse the light source to balance the charge from the detector crystal (this only works if the detector signal charge is negative, for an N-channel FET; you could use a P-channel device otherwise, but their properties are inferior).
They got huge noise reductions in their preamp and attributed this to the elimination of the feedback resistor. But as the sort of thing that often happened in the pre-Ma Bell breakup days, a couple of smarty-pants (Kern and Mackenzie if memory serves) from Bell Labs pointed out that the primary improvement was due to their decapsulation of the JFET and the reduction in thermal noise from the lossy borosilicate sealing glass package dielectric. K&M went on to prove this by getting some FETs in alumina or boron nitride or beryllium oxide headers and using a big feedback R as was conventional, and got about the same results. But most of the Nuke Sci folks remembered the opto-reset thing because it sounded so cool.
As far as the LS627 though---the benefits of photocontrol seem to be analogous to the real estate adage: isolation, isolation, and isolation. As PRR indicates, you can get rid of most of the second harmonic distortion with the feedback of half of drain voltage to the gate---but then what's the advantage of the isolation? If it's really just the single JFET shown, in order to keep the device off you need some connection to the gate to apply a negative bias. This has to be a big resistor or you kill your photosensitivity. But if large you will get distortion from the modulation of the channel-gate capacitance (coincidentally, I just saw this effect while making some improvements to a client's DMOS-FET based limiter---they were using a 2N7000, which to my surprise did fairly well in the app).
Where the photogeneration of drain-gate leakage shines (sorry couldn't resist) is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor. Landis et al. did this at Lawrence Berkeley in the 70's iirc. You can have as your only feedback element a tiny capacitor, and periodically pulse the light source to balance the charge from the detector crystal (this only works if the detector signal charge is negative, for an N-channel FET; you could use a P-channel device otherwise, but their properties are inferior).
They got huge noise reductions in their preamp and attributed this to the elimination of the feedback resistor. But as the sort of thing that often happened in the pre-Ma Bell breakup days, a couple of smarty-pants (Kern and Mackenzie if memory serves) from Bell Labs pointed out that the primary improvement was due to their decapsulation of the JFET and the reduction in thermal noise from the lossy borosilicate sealing glass package dielectric. K&M went on to prove this by getting some FETs in alumina or boron nitride or beryllium oxide headers and using a big feedback R as was conventional, and got about the same results. But most of the Nuke Sci folks remembered the opto-reset thing because it sounded so cool.
Svart
Well-known member
not to totally take this thread off topic but have you checked out the new photoMOS stuff?
Pretty cool to have an optoisolated MOSfet.
Pretty cool to have an optoisolated MOSfet.
mikep
Well-known member
yeah, I don't know what the point would be. I didnt give it too much thought before posting. I did realize that the gate connection would be useful for distortion reduction. I guess I was thinking it might "sound different".
[quote author="bcarso"]...is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor.[/quote]
wow. I know exactly what you are referring to. I was reading about gamma ray detector preamps just yesterday. when I saw those FETS I was looking for spice models on the linear sys website and was firmly thinking in audio-mode. I didn't conciously put it together that these were the same type of part until you mentioned it. FWIW, nowadays they use a BJT transistor switch to reset the integrator, which is driven by the photo-fet. here's a manufacturer of this stuff: canberra.com
mike p
[quote author="bcarso"]...is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor.[/quote]
wow. I know exactly what you are referring to. I was reading about gamma ray detector preamps just yesterday. when I saw those FETS I was looking for spice models on the linear sys website and was firmly thinking in audio-mode. I didn't conciously put it together that these were the same type of part until you mentioned it. FWIW, nowadays they use a BJT transistor switch to reset the integrator, which is driven by the photo-fet. here's a manufacturer of this stuff: canberra.com
mike p
bcarso
Well-known member
[quote author="mikep"]yeah, I don't know what the point would be. I didnt give it too much thought before posting. I did realize that the gate connection would be useful for distortion reduction. I guess I was thinking it might "sound different".
[quote author="bcarso"]...is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor.[/quote]
wow. I know exactly what you are referring to. I was reading about gamma ray detector preamps just yesterday. when I saw those FETS I was looking for spice models on the linear sys website and was firmly thinking in audio-mode. I didn't conciously put it together that these were the same type of part until you mentioned it. FWIW, nowadays they use a BJT transistor switch to reset the integrator, which is driven by the photo-fet. here's a manufacturer of this stuff: canberra.com
mike p[/quote]
I wound up using a bipolar myself in dual cooled preamps for a Reticon photodiode array (this was in 1978). It was resetting the charge on a 150fF cap, and it was very tricky to control to avoid slamming the preamp output into the rail, as one can imagine.
The bipolar was a small-geometry RF transistor (MRF502) which I selected for low Icbo at room temperature; some of them were as low as 100fA with 5V bias, and this went down with the cooling.
EDIT: the paper describing this and other aspects of the system can be found in an old SPIE proceedings from a conference called Instrumentation in Astronomy IV, circa 1982. I have some hard-copy reprints somewhere if anyone is interested, and I'll correct the typos if I can remember them all, particularly the ones in the equations. I haven't checked to see if SPIE has scanned it into their online database yet---probably not. The proceedings were scarce the last time I looked on bookfinder.
When I was just finishing the system---perilously close to being fired for taking so long---my friend told her friend, the late author F. Israel Regardie, about it, and he reportedly said "Well let's hope that this is not the last thing he does." I know exactly what he meant now, years later. Although I've done plenty of other stuff, this does represent something of a pinnacle so far.
[quote author="bcarso"]...is in super-low-current applications, for example providing a resetting charge dispenser for x-ray detector preamps in lieu of a feedback resistor.[/quote]
wow. I know exactly what you are referring to. I was reading about gamma ray detector preamps just yesterday. when I saw those FETS I was looking for spice models on the linear sys website and was firmly thinking in audio-mode. I didn't conciously put it together that these were the same type of part until you mentioned it. FWIW, nowadays they use a BJT transistor switch to reset the integrator, which is driven by the photo-fet. here's a manufacturer of this stuff: canberra.com
mike p[/quote]
I wound up using a bipolar myself in dual cooled preamps for a Reticon photodiode array (this was in 1978). It was resetting the charge on a 150fF cap, and it was very tricky to control to avoid slamming the preamp output into the rail, as one can imagine.
The bipolar was a small-geometry RF transistor (MRF502) which I selected for low Icbo at room temperature; some of them were as low as 100fA with 5V bias, and this went down with the cooling.
EDIT: the paper describing this and other aspects of the system can be found in an old SPIE proceedings from a conference called Instrumentation in Astronomy IV, circa 1982. I have some hard-copy reprints somewhere if anyone is interested, and I'll correct the typos if I can remember them all, particularly the ones in the equations. I haven't checked to see if SPIE has scanned it into their online database yet---probably not. The proceedings were scarce the last time I looked on bookfinder.
When I was just finishing the system---perilously close to being fired for taking so long---my friend told her friend, the late author F. Israel Regardie, about it, and he reportedly said "Well let's hope that this is not the last thing he does." I know exactly what he meant now, years later. Although I've done plenty of other stuff, this does represent something of a pinnacle so far.
Similar threads
