I want to clamp the input voltage of my linear amp board to +- 1.6 v with a led. I have some Lumex LX3059srsgw led. the data sheet doesn't show anything like that. Power dissipation and a 30 ma current limit. They are bidirectional power supply red and green so they would clamp voltage 2.0 v either way.
I have a pi filter on amplifier input: 10 uf cap in series with 2 ea 4.7k resistors, then the input transistor base. 27 pf cap is between 4.7k resistor center and input transistor emitter. Max amp output supposed to be at 1.5 VAC. I want to parallel the 27 pf cap with the led. I don't know the capacitance. 6 ea 1n4148 would do it, 3 back to back either way but I don't have room. 1n4148 are showing 4 pf capacitance. Back to back bzx79-2v4 zeners would be close, clamping at 3 v, but those have 450 pf of off capacitance.
So does anybody have a red/green two lead low power led and a capacitance meter that goes below 100 pf? Can you find out the off state capacitance ? thanks.
I have a pi filter on amplifier input: 10 uf cap in series with 2 ea 4.7k resistors, then the input transistor base. 27 pf cap is between 4.7k resistor center and input transistor emitter. Max amp output supposed to be at 1.5 VAC. I want to parallel the 27 pf cap with the led. I don't know the capacitance. 6 ea 1n4148 would do it, 3 back to back either way but I don't have room. 1n4148 are showing 4 pf capacitance. Back to back bzx79-2v4 zeners would be close, clamping at 3 v, but those have 450 pf of off capacitance.
So does anybody have a red/green two lead low power led and a capacitance meter that goes below 100 pf? Can you find out the off state capacitance ? thanks.
Last edited:
LX3059srsgw is a 3 lead device, common cathode according to data sheet ?
Looks like up to a 0.5V difference in forward voltage for the Red/ Green diodes.
This will not work for even handed clamping ?
Looks like up to a 0.5V difference in forward voltage for the Red/ Green diodes.
This will not work for even handed clamping ?
The transistor has back voltage Vbe of 6. I just don't want to blow anything up if I turn the disco mixer up too loud- it could put out 6 VAC pretty easily. I can parallel the two LED outer (hot) leads pretty easily. the output of the amp will clip somewhere between 1.5 v and 2, I don't have to worry about the sound, it would be clipping badly with this much input.
Last edited:
The common-cathode LED won't be bidirectional. Only half the AC wave will be clamped. You want the two-lead "tricolor" type of LED.
I don't want to wait a week and pay $6 freight. Actually, since the transistor conducts forwards, I could use one led backwards across b-e. The armstrong 621amp used one silicon diode backwards across the input be junction, I don't see the sonic wisdom of this unless they selected for 0.7 v diodes and 0.6 v transistors. But a led doesn't conduct until way above 0.6 v. It should be inert at 1.5 vac full power input except for capacitance.
Does anybody know the capacitance of red low power LED's? I've got some old HP datasheets and they never specified dark capacitance.
Does anybody know the capacitance of red low power LED's? I've got some old HP datasheets and they never specified dark capacitance.
Last edited:
The capacitance could well be dynamic, dependent on applied voltage and polarity. A Googled EDN article stated a typical LED capacitance of 50pF. Perhaps doubled with paralleled pn junctions.
Last edited:
50 pf thank you. I will just leave out the 27 pf cap on the print (a non-linear x7u) and put in a non-linear red led the opposite way from the b-e junction. Had it been 450 pf like a zener, that might have been a sound problem.
Like any PN junction the LED capacitance WILL depend on the applied voltage - maximum capacitance as the diode starts to break into forward conduction, decreasing as the REVERSE bias voltage increases. For practical purposes there is little change in the capacitance for reverse voltages beyond about 3 - 4 volts.
I believe the basic device physics gives a pretty accurate prediction of the junction capacitance, so a decent SPICE model of the LED could be used to determine the capacitance-versus-reverse bias curve in simulation. (Unfortunately, the one time I tried to verify an LED's published SPICE model against measured behavior the agreement was rather poor but, of course, that doesn't prove ALL models for LED's are bad.)
Is it the article "An LED's intrinsic capacitance works in a 650-mV LRC circuit" at < http://www.edn.com/design/led/4368392/An-LED-s-intrinsic-capacitance-works-in-a-650-mV-LRC-circuit >? It doesn't give any reference for the "typical" value they quote, but the circuit described in the article seems to suggest some clever ideas for measuring the capacitance.
Dale
Philips publishes that data: for the CQT24 (an ordinary, 5mm bicolor LED) the typical capacitance @0V & 1MHz is 160pF
As the LED starts to conduct the amp should be way into the clipping realm, so sounding stupid with non-linear capacitance is not a problem when the LED is protecting the input transistor. I just hope the LED is sonicly invisible in the dark zone where the amp should be performing accurately. I assume the function of the 27 pf cap is to prevent RF from flowing through the amp.
I'm economically prohibited from using SPICE since it doesn't operate on Linux systems. If I could budget for $100 Windows XP, microsoft doesn't sell it anymore. Up to date op systems like ubuntu13 lock this computer up totally with extraneous features. I assume Win 8,9,10 would be as bad or worse.
Some LEDs capacitance have been specified for use in TosLink type applications
LTSpice works very well on Ubuntu under Wine and in the past I had a Linux version of the free light edition of Simetrix
LTSpice works very well on Ubuntu under Wine and in the past I had a Linux version of the free light edition of Simetrix
Those capacitances sound very high, especially compared to diodes designed for ESD protection and the like. You could put two or three in series to get more than 0.6V, and that'd reduce the capacitance even further. You'd end up with a couple of pF instead of 20-50pF.
Yes… In case I was unclear, you could use traditional Si junctions instead of LEDs, but put a few in series to get the higher threshold voltage that seems to be the point of using an LED. In this way, you would not have to tolerate the huge capacitance, or the photosensitivity that may cause spurious signals to be injected into the circuit. It seems likely that if the LED was not painted black, it could act as a photo receiver and inject noise into the circuit from pulsed light sources such as fluorescent lamps or modern HF driven LED lamps.
It may be worthwhile to review the old Burr-Brown Ap Note, "Diode Connected FET Protects Op Amps" at http://www.ti.com/lit/an/sboa058/sboa058.pdf .
Dale
Limiting the input voltage to 0.6 v with a jfet as in this ap note cuts off those legitimate signals that peak at at 1.5 v, the spec for maximum power out of the amp. The input transistor, a MPS8099 (formerly BC108C), allows much more current than a fet input op amp. Three 1n4148 in series is a suitable electrical replacement to prevent 5 v backwards on the transistor be junction, but is a clumsy kluge when building by twisting the leads together and soldering. The blank board is already drilled and loaded with parts: I have room for a 2 leg device, the 27 pf ceramic disk cap. I'll take a black marks-a-lot and paint the LED black as suggested, and put it in instead of the disk cap. Maybe a green LED that conducts at ~2v has even less non-linear effect at 1.5 v normal signal in than a 1.6 v RED led.
- Status
- Not open for further replies.