Here is another mod I whipped up for the O2, a clipping detect circuit for the first (gain) stage.
Note that if you know exactly what your source output RMS voltage level is then this circuit is pointless since you can calculate the gains you need to avoid clipping from the 7Vrms max (AC) and 4.5Vrms max (batteries) formulas that RocketScientist has in the writeup. This circuit is for situations where you don't know exactly what the output voltage level of your source is as the O2 is carried from place to place and source to source. The circuit will tell whether any distortion being heard is clipping or not.
This circuit is also untested at this point, although I will be building this one as time permits over the next week or two. The circuit is based on this one from the net
Audio Clipping Indicator - RED - Page132
The design uses the NJM2901 comparator, the quad version of the NJM2903 comparator RocketScientist is using in the O2.
NJM2901N NJR Comparator ICs
Each pair of comparators on each channel form a "window comparator" that triggers if the peak output voltage level of that channel at any instant in time exceeds an upper or lower voltage reference point. For the reference points I'm using a string of 1N4148 diodes from each rail that set the upper level at 2.1v below the top rail, wherever it is at the moment, and the bottom point at 2.1V above the bottom rail. This way the reference points "float" with the rails, so that as the batteries run down the comparator window reference points drop along with them (the comparator window narrows), always staying 2.1V away. It should be possible to use red LEDs with a Vf = 2.2V to replace the diodes strings.
Once the comparator triggers then pulse is lengthened enough to be able to see (a tiny flash wouldn't be visible), then turns on an open collector transistor that powers a LED. The LED would go out to the front or rear panels.
The second channel works the same way with the open collector transistors doing a logical "OR" between the two for the single output LED. If desired each channel could run its own LED so it would be visible which channel is clipping.
The inputs should be picked up from the output of U1 on the O2 board (pins 1 and 7), which is the same as the inputs to the pot. The best places to connect are probably the end of R16 and R22 that connects to the pot. V+, V- and ground and be picked up at any convenient spot after the power management circuit.
To test, disconnect the headphones, set the O2 at the highest gain position (assume 6.5x here) and turn the source all the way up. The position of the pot doesn't matter since this circuit detects clipping in the first stage - before the pot. If the clipping LED does not light then congratulations, your source voltage level multiplied by the 6.5 gain setting is less than the first stage clipping level! If the LED does light then you are clipping. Flip the gain switch to the lower position and it should stop. If not then you need to reduce the gain resistors in that lowest setting, down from the stock 2.5x, until the clipping light does not light anymore.
If the light did not light even in the 6.5x gain position you can install resistors for higher gains (7x, 10x) until the clipping LED does light to test it. Then chose the resistors for the highest gain setting that did not light the clipping LED.
In the sim plot below the batteries are already assumed to be run down and asymetrical. The positive battery (rail) is 8.4V while the negative battery (rail) still has more charge at 9V. This will be the typical situation - the two batteries will never be exactly equal of course. The weakest battery, with the lowest voltage, will be the one the amp hits first and clips on. So in this case if all is working correctly the clipping should occur on the positive 8.4V rail first and trigger the circuit.
The plot shows that is exactly what happens. Then 60mS along a second voltage source turns on and bucks the first one by 0.2V, enough to barely drop the amp level below clipping, and the clipping indicator then turns off. One plot is done at 200Hz and the second at 2kHz.
The circuit could be mounted on an upper PC board that would slide into the top slot on the B3-080 chasisd, or it may be possible to build it up as a small daughter board and mount it over the O2 board on something like a nylon bolt - that is what I'm going to try anyway
- then mount the indicator led on the front panel with thermal glue.
Note that this circuit does not detect the actual clipping! Rather it detects when the output voltage level on either channel of the O2's first stage exceeds a voltage level known to cause clipping - that 7Vrms for AC and 4.5Vrms for batteries that RocketScientist has mentioned in the literature. The circuit is making the assumption that if the output level exceeds the level that causes clippiing, then clipping will be occurring.
Note that if you know exactly what your source output RMS voltage level is then this circuit is pointless since you can calculate the gains you need to avoid clipping from the 7Vrms max (AC) and 4.5Vrms max (batteries) formulas that RocketScientist has in the writeup. This circuit is for situations where you don't know exactly what the output voltage level of your source is as the O2 is carried from place to place and source to source. The circuit will tell whether any distortion being heard is clipping or not.
This circuit is also untested at this point, although I will be building this one as time permits over the next week or two. The circuit is based on this one from the net
Audio Clipping Indicator - RED - Page132
The design uses the NJM2901 comparator, the quad version of the NJM2903 comparator RocketScientist is using in the O2.
NJM2901N NJR Comparator ICs
Each pair of comparators on each channel form a "window comparator" that triggers if the peak output voltage level of that channel at any instant in time exceeds an upper or lower voltage reference point. For the reference points I'm using a string of 1N4148 diodes from each rail that set the upper level at 2.1v below the top rail, wherever it is at the moment, and the bottom point at 2.1V above the bottom rail. This way the reference points "float" with the rails, so that as the batteries run down the comparator window reference points drop along with them (the comparator window narrows), always staying 2.1V away. It should be possible to use red LEDs with a Vf = 2.2V to replace the diodes strings.
Once the comparator triggers then pulse is lengthened enough to be able to see (a tiny flash wouldn't be visible), then turns on an open collector transistor that powers a LED. The LED would go out to the front or rear panels.
The second channel works the same way with the open collector transistors doing a logical "OR" between the two for the single output LED. If desired each channel could run its own LED so it would be visible which channel is clipping.
The inputs should be picked up from the output of U1 on the O2 board (pins 1 and 7), which is the same as the inputs to the pot. The best places to connect are probably the end of R16 and R22 that connects to the pot. V+, V- and ground and be picked up at any convenient spot after the power management circuit.
To test, disconnect the headphones, set the O2 at the highest gain position (assume 6.5x here) and turn the source all the way up. The position of the pot doesn't matter since this circuit detects clipping in the first stage - before the pot. If the clipping LED does not light then congratulations, your source voltage level multiplied by the 6.5 gain setting is less than the first stage clipping level! If the LED does light then you are clipping. Flip the gain switch to the lower position and it should stop. If not then you need to reduce the gain resistors in that lowest setting, down from the stock 2.5x, until the clipping light does not light anymore.
If the light did not light even in the 6.5x gain position you can install resistors for higher gains (7x, 10x) until the clipping LED does light to test it. Then chose the resistors for the highest gain setting that did not light the clipping LED.
In the sim plot below the batteries are already assumed to be run down and asymetrical. The positive battery (rail) is 8.4V while the negative battery (rail) still has more charge at 9V. This will be the typical situation - the two batteries will never be exactly equal of course. The weakest battery, with the lowest voltage, will be the one the amp hits first and clips on. So in this case if all is working correctly the clipping should occur on the positive 8.4V rail first and trigger the circuit.
The plot shows that is exactly what happens. Then 60mS along a second voltage source turns on and bucks the first one by 0.2V, enough to barely drop the amp level below clipping, and the clipping indicator then turns off. One plot is done at 200Hz and the second at 2kHz.
The circuit could be mounted on an upper PC board that would slide into the top slot on the B3-080 chasisd, or it may be possible to build it up as a small daughter board and mount it over the O2 board on something like a nylon bolt - that is what I'm going to try anyway
- then mount the indicator led on the front panel with thermal glue.Note that this circuit does not detect the actual clipping! Rather it detects when the output voltage level on either channel of the O2's first stage exceeds a voltage level known to cause clipping - that 7Vrms for AC and 4.5Vrms for batteries that RocketScientist has mentioned in the literature. The circuit is making the assumption that if the output level exceeds the level that causes clippiing, then clipping will be occurring.
Attachments
Last edited:
Interesting! I'll take a look at that. Adding a Schottky like a 1N60P would bump it up by 0.2V.
Thanks for the heads up!
Attached is an update on the O2 clip detect circuit. Just a reminder this circuit is absolutely not needed in the majority of cases with the O2. As RocketScientist has written several times the vast majority of sources should be within the default gain resistor settings for the O2. For the few remaining higher voltage sources if the level is known the resistors can be calculated from the O2 instructions.
This circuit only really has use in cases where te O2 might be taken on the road for use with many different sources in situations where their output levels are not known and might contain the few with higher output voltages. That coupled with an expected need to use the higher gain setting, more likely to clip, due to some source material recorded at lower-than-normal volume levels.
* The IC is changed to the TL064 quad opamp used in the original circuit, from the quad NJM quad comparator, to eliminate the need for the 4 pullup resistors and associated idle current draw. The TL064 has low quiescent current draw and has a slew rate adequate for up to the 20kHz rate triggers. I tried several NJM quad op amps including the NJM2902 and NJM3403A. I found that the only one with a high enough slew rate at 20kHz was two of the NJM2068 used in the O2 gain stage. but two of those would have considerably higher quiescent current than the TL064. Low distortion and low noise, features of the NJM2068, don't matter here for use as a comparator.
The TL064 is shown as individual op amps in the simulation circuit below due to TI only supplying the TL064 sim model as 1/4 of the total chip. All 4 of those really are on the same quad chip.
TL064BCN Texas Instruments Op Amps
* The window reference resistor is increased to 100K from 20k to further reduce current draw. The TL064 has jfet inputs that are still less than 1/100 the current through the 100K.
* 1uF MLCC ceramic bypass capacitor added across the 100K window reference resistor. The cap removes small voltage spikes caused by the resulting IR drop from input bias current from the op amps.
* The two output diodes changed to Schottky. Any small signal Schottky diode good for at least 10mA will work here.
* 0.22uF decoupling capacitor added across the rails, to be mounted near the op amp, given than this board may be a daughter board mounted above the O2 board fed by long power supply leads to the board.
* It should be entirely possible to replace the 4 1N4148s with one 3mm red LED with a Vf of 2.2V. The Vf of any given LED would need to be measured at the bias current of around 0.13mA. It may be necessary to add one 1N4148 in series with the LED to get to 2.2V at that low current level.
Plots below show the clipping indicator triggering with a 200Hz signal and a 20kHz signal. I've reversed the order of events over the previous plots. The circuit starts out with the input audio level slightly less than the window trigger points (clipping LED off), then 0.2V is added to that at 50mS to push it above the upper trigger point (clipping LED on). The aqua line in both plots is the collector of the transistors. When it goes low the clipping LED turns on.
This circuit only really has use in cases where te O2 might be taken on the road for use with many different sources in situations where their output levels are not known and might contain the few with higher output voltages. That coupled with an expected need to use the higher gain setting, more likely to clip, due to some source material recorded at lower-than-normal volume levels.
* The IC is changed to the TL064 quad opamp used in the original circuit, from the quad NJM quad comparator, to eliminate the need for the 4 pullup resistors and associated idle current draw. The TL064 has low quiescent current draw and has a slew rate adequate for up to the 20kHz rate triggers. I tried several NJM quad op amps including the NJM2902 and NJM3403A. I found that the only one with a high enough slew rate at 20kHz was two of the NJM2068 used in the O2 gain stage. but two of those would have considerably higher quiescent current than the TL064. Low distortion and low noise, features of the NJM2068, don't matter here for use as a comparator.
The TL064 is shown as individual op amps in the simulation circuit below due to TI only supplying the TL064 sim model as 1/4 of the total chip. All 4 of those really are on the same quad chip.
TL064BCN Texas Instruments Op Amps
* The window reference resistor is increased to 100K from 20k to further reduce current draw. The TL064 has jfet inputs that are still less than 1/100 the current through the 100K.
* 1uF MLCC ceramic bypass capacitor added across the 100K window reference resistor. The cap removes small voltage spikes caused by the resulting IR drop from input bias current from the op amps.
* The two output diodes changed to Schottky. Any small signal Schottky diode good for at least 10mA will work here.
* 0.22uF decoupling capacitor added across the rails, to be mounted near the op amp, given than this board may be a daughter board mounted above the O2 board fed by long power supply leads to the board.
* It should be entirely possible to replace the 4 1N4148s with one 3mm red LED with a Vf of 2.2V. The Vf of any given LED would need to be measured at the bias current of around 0.13mA. It may be necessary to add one 1N4148 in series with the LED to get to 2.2V at that low current level.
Plots below show the clipping indicator triggering with a 200Hz signal and a 20kHz signal. I've reversed the order of events over the previous plots. The circuit starts out with the input audio level slightly less than the window trigger points (clipping LED off), then 0.2V is added to that at 50mS to push it above the upper trigger point (clipping LED on). The aqua line in both plots is the collector of the transistors. When it goes low the clipping LED turns on.
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.