Changing those components is not the right way to do it, because at best they will give you a 6dB/octave rolloff. The conventional way is to use an active filter at line level. But before going too far down this path of discussion: what are you trying to achieve? What is the application?
hi shawn, thanks for the post.
I may be ackward with my question as I do not have a electrical science background.
I'm working on a bi amplification project.and the L12-2 board picked my interest a couple of years ago. I have a minidsp on hand in order to do the split of the audio frequencies between bass amp and mid/high amp.
I also did consider active 2 or 3 way crossover board like eliott sound project 09 (http://sound.whsites.net/project09.htm)
the other day, I came accross this page on the trans-f audio website:
https://www.trans-fi.com/my-system
I will quote:
The PRV full range drivers are powered by the new Temple Audio Mono2 amps (...)
This amp has the added feature of a user replaceable input cap which acts as a selectable high pass filter stopping unecessary low frequencies (already handled by the woofer) from reaching the Full Range driver. This has a dramatic effect of cleaning up the sound, I think mainly because the amp isn't having to work so hard reproducing the lower frequencies.
One the back are outputs linked directly to the volume control to drive the sub amp.
Then remembering the page of calvin on those L12-2 Board, with the input going though a band pass filter, I was left wondering if it was possible to modulate this band pass filter directly on board in order to achieve kind of the same effect: releasing the board from reproducing the 20-300Hz even if fed by the full audio spectrum (20HZ- 20KHz)
Is it now clearer to you?
I may be ackward with my question as I do not have a electrical science background.
I'm working on a bi amplification project.and the L12-2 board picked my interest a couple of years ago. I have a minidsp on hand in order to do the split of the audio frequencies between bass amp and mid/high amp.
I also did consider active 2 or 3 way crossover board like eliott sound project 09 (http://sound.whsites.net/project09.htm)
the other day, I came accross this page on the trans-f audio website:
https://www.trans-fi.com/my-system
I will quote:
The PRV full range drivers are powered by the new Temple Audio Mono2 amps (...)
This amp has the added feature of a user replaceable input cap which acts as a selectable high pass filter stopping unecessary low frequencies (already handled by the woofer) from reaching the Full Range driver. This has a dramatic effect of cleaning up the sound, I think mainly because the amp isn't having to work so hard reproducing the lower frequencies.
One the back are outputs linked directly to the volume control to drive the sub amp.
Then remembering the page of calvin on those L12-2 Board, with the input going though a band pass filter, I was left wondering if it was possible to modulate this band pass filter directly on board in order to achieve kind of the same effect: releasing the board from reproducing the 20-300Hz even if fed by the full audio spectrum (20HZ- 20KHz)
Is it now clearer to you?
No problemedit: shaun (sorry I mispelled your name)
Quite off topic now, but your miniDSP can do everything you need.
yes I guess so... as long as you stick with the miniDSP internal DAC..
I have a couple of friends that would be nice enough to lend my their dacs to try out in my system. That's why I was looking for an other way to do the same thing , but with a regular dac. my bass amp board as a dsp built in, so feeding him with a full spectrum, he will apply the low pass filter to keep only the sub 300Hz,
But my main amp (300Hz and Up) will still reproduce the full spectrum including the sub 300Hz.
I guess they implemented something like an opamp high pass filter at the input of their amp?
Any suggestion to easily do so?
Best regards
I have a couple of friends that would be nice enough to lend my their dacs to try out in my system. That's why I was looking for an other way to do the same thing , but with a regular dac. my bass amp board as a dsp built in, so feeding him with a full spectrum, he will apply the low pass filter to keep only the sub 300Hz,
But my main amp (300Hz and Up) will still reproduce the full spectrum including the sub 300Hz.
I guess they implemented something like an opamp high pass filter at the input of their amp?
Any suggestion to easily do so?
Best regards
I guess something like that might do the trick ?
https://www.ebay.fr/itm/AUDIO-VLF-H...h=item1a2c7fecf7:g:YSIAAOSwdzVXkJr1:rk:1:pf:0
https://www.ebay.fr/itm/AUDIO-VLF-H...h=item1a2c7fecf7:g:YSIAAOSwdzVXkJr1:rk:1:pf:0
LJM is the commercial designer and producer of the amplifier modules and kits. It's proper that he doesn't use the forum to comment on or compare other manufacturer's designs and products. He does offer one (LM3886) module but for all Chipamp matters, the right forum is here: https://www.diyaudio.com/forums/chip-amps/
Hi,
#307
better is such a relative word
I just know that the TDA that works as amplifier in my DAAS32 measurement system has blown twice at still low levels when I tried to measure my electrostats.
It simply doesn´t like to drive complex loads.
The L12-2 on the other hand is a very stable amplifier already without a Boucherot cell at its´s output.
I use it as measurement amplifier since then and never had the slightest problem.
As I wrote on my website it also keeps sonically sober when driving complex loads.
So only if extreme compactness of the build were the primary goal, I´d consider the TDA ... all else I´d chose the L12-2 hands down.
jauu
Calvin
#307
better is such a relative word
I just know that the TDA that works as amplifier in my DAAS32 measurement system has blown twice at still low levels when I tried to measure my electrostats.
It simply doesn´t like to drive complex loads.
The L12-2 on the other hand is a very stable amplifier already without a Boucherot cell at its´s output.
I use it as measurement amplifier since then and never had the slightest problem.
As I wrote on my website it also keeps sonically sober when driving complex loads.
So only if extreme compactness of the build were the primary goal, I´d consider the TDA ... all else I´d chose the L12-2 hands down.
jauu
Calvin
L12 Noise Question
I have a pair of L12s and one seems to have noise that is not present in the other. The good amp works fine and is very quiet as I vary my input attenuator from max attenuation to zero attenuation with no input signal. The problematic amp has a noticeably higher noise under the same scenario, and curiously, the noise is worse at max input attenuation. I'm wondering where in the L12 circuit I should be troubleshooting. Because the noise decreases as I lower the input attenuation and with no input signal present, I suspect something in the front end/diff amp portion of the L12. Looking for ideas. The amps came already assembled but have sat on my shelf for about 2 years, I only recently fired them up. Thank you for any advice you can offer.
I have a pair of L12s and one seems to have noise that is not present in the other. The good amp works fine and is very quiet as I vary my input attenuator from max attenuation to zero attenuation with no input signal. The problematic amp has a noticeably higher noise under the same scenario, and curiously, the noise is worse at max input attenuation. I'm wondering where in the L12 circuit I should be troubleshooting. Because the noise decreases as I lower the input attenuation and with no input signal present, I suspect something in the front end/diff amp portion of the L12. Looking for ideas. The amps came already assembled but have sat on my shelf for about 2 years, I only recently fired them up. Thank you for any advice you can offer.
I haven't had such a problem with either of 2 pairs of original, prebuilt L12/2 modules by LJM but first, I would check that my source and preamp were OK by swapping inputs to either amplifier and verifying that the fault remains with the same amp. channel.
Noise can be difficult to isolate since it is dynamic, like oscillation and the direct coupled circuit of most linear audio amplifiers has feedback which means noise or DC problems can arise almost anywhere but seem to manifest in different locations. I would begin with a physical check by placing a finger on semis, beginning with the diffamp pair and listening for any change. A little body heat should alter any noise if it's source is a bad semi. A little pressure on other components should also make a difference if other passive components are involved but remember your safety requirements! >30V across your body should be considered unsafe so external heat (perhaps a hair dryer with nozzle attached) can be used to warm up just a part of the board to help isolate the area of any fault. This could be soldering, a PCB track break or even a bad passive component.
A dual bench power supply is a great asset for reducing supply voltages to safe limits. I use one for the rare occasions when I have such a problem that requires poking about whilst the amp. is powered up.
You can identify faulty resistors and semis by making voltage checks at relevant circuit nodes (follow a correct schematic for this) and compare with the good channel. This can be tedious but isolating the area reduces the time substantially. If this draws a blank, move on to those caps, though you seldom find noise problems with ceramics, film or electrolytics as fitted by LJM. I can't speak for assemblies from other manufacturers though.
If all else fails yet the semis prove to be fine, remove the PCB assembly from the heatsink and reflow the solder joints, starting with the input stage.
I hope this helps or at least gives you an idea of where to start.
Noise can be difficult to isolate since it is dynamic, like oscillation and the direct coupled circuit of most linear audio amplifiers has feedback which means noise or DC problems can arise almost anywhere but seem to manifest in different locations. I would begin with a physical check by placing a finger on semis, beginning with the diffamp pair and listening for any change. A little body heat should alter any noise if it's source is a bad semi. A little pressure on other components should also make a difference if other passive components are involved but remember your safety requirements! >30V across your body should be considered unsafe so external heat (perhaps a hair dryer with nozzle attached) can be used to warm up just a part of the board to help isolate the area of any fault. This could be soldering, a PCB track break or even a bad passive component.
A dual bench power supply is a great asset for reducing supply voltages to safe limits. I use one for the rare occasions when I have such a problem that requires poking about whilst the amp. is powered up.
You can identify faulty resistors and semis by making voltage checks at relevant circuit nodes (follow a correct schematic for this) and compare with the good channel. This can be tedious but isolating the area reduces the time substantially. If this draws a blank, move on to those caps, though you seldom find noise problems with ceramics, film or electrolytics as fitted by LJM. I can't speak for assemblies from other manufacturers though.
If all else fails yet the semis prove to be fine, remove the PCB assembly from the heatsink and reflow the solder joints, starting with the input stage.
I hope this helps or at least gives you an idea of where to start.
L12-2 idle noise
r.e. #314
Thanks Ian for your prompt and detailed reply. I have already done lots of swapping input and output paths with the amps in their chassis, including even the speaker protection relays. I'm pretty sure the problem is in the amp itself, and that the two amps are performing differently.
FYI these amps are retrofitted into a salvaged Sony 55ES chassis, which provides a robust DC supply of 56V. (The Sony's STK3102 died and I have given up trying to find a non-counterfeit replacement). It appears that the L12 caps are rated at 63V so from the beginning I presumed I had enough margin operating at 56VDC, what do you think? I note from the paperwork when I originally purchased the L12s that they were spec'd up to 55 VDC but the latest offerings on eBay spec the supply only at 50V. Is it possible that I am too close to the margin? Could a noise at idle problem be caused by running the VDC too high?
I have a variac and a dual 30v supply that I can use for troubleshooting. Because of the way the amps are mounted in the Sony it is difficult for me to take individual measurements on the L12 components so before removing the amps from the Sony (again, I have done a lot of signal path swapping so for now I'm pretty sure the amps are performing differently) I'm thinking of running the system up on the variac and seeing if I can detect a Sony VDC supply voltage where the noise begins. I'm also thinking of letting the amps warm up for a few hours and see if an extended warmup period has any effect of the noise.
After that if the noise is present at VDC<30 with the amps installed in the Sony chassis I will then remove the amps from the Sony chassis and use the benchtop supply to explore further any circuit measurement differences I can see up to 30VDC. If I make it this far I will focus on the components and voltage measurements before the VAS in addition to using your suggestions for poking, prodding and perhaps ultimately touching up solder joints, which all make sense to me.
Thanks again for your advice and I will report back what happens next after spending some more time on this.
skipt
r.e. #314
Thanks Ian for your prompt and detailed reply. I have already done lots of swapping input and output paths with the amps in their chassis, including even the speaker protection relays. I'm pretty sure the problem is in the amp itself, and that the two amps are performing differently.
FYI these amps are retrofitted into a salvaged Sony 55ES chassis, which provides a robust DC supply of 56V. (The Sony's STK3102 died and I have given up trying to find a non-counterfeit replacement). It appears that the L12 caps are rated at 63V so from the beginning I presumed I had enough margin operating at 56VDC, what do you think? I note from the paperwork when I originally purchased the L12s that they were spec'd up to 55 VDC but the latest offerings on eBay spec the supply only at 50V. Is it possible that I am too close to the margin? Could a noise at idle problem be caused by running the VDC too high?
I have a variac and a dual 30v supply that I can use for troubleshooting. Because of the way the amps are mounted in the Sony it is difficult for me to take individual measurements on the L12 components so before removing the amps from the Sony (again, I have done a lot of signal path swapping so for now I'm pretty sure the amps are performing differently) I'm thinking of running the system up on the variac and seeing if I can detect a Sony VDC supply voltage where the noise begins. I'm also thinking of letting the amps warm up for a few hours and see if an extended warmup period has any effect of the noise.
After that if the noise is present at VDC<30 with the amps installed in the Sony chassis I will then remove the amps from the Sony chassis and use the benchtop supply to explore further any circuit measurement differences I can see up to 30VDC. If I make it this far I will focus on the components and voltage measurements before the VAS in addition to using your suggestions for poking, prodding and perhaps ultimately touching up solder joints, which all make sense to me.
Thanks again for your advice and I will report back what happens next after spending some more time on this.
skipt
Good that you have already covered the general fault elimination steps. I agree with your concern though, that you are also pushing the safe limits of the component ratings and power. An unregulated +/- 56V supply is above the recommended +20% safety buffer for 63V rated caps, regarding supply surges and spikes and its likely that the TO92 semis aren't original parts. This could be a problem, which is why I would be checking the operation of these parts with a little extra heat and prodding before getting down to more invasive methods.
Early in the thread and on his website (see #311), Calvin emphasised the supply limits he found necessary and of course, +/-50V max is recommended by LJM. Whilst I have heard odd noises like faint hum and crackling when the voltage supply is too high for many audio circuits, it doesn't necessarily follow that this is the problem here.
As a more general comment, Japanese manufacturers often used relatively small power transformers with high rail voltages. These have poor regulation, so the supply sags under heavy, low impedance loads. This automatically gives some protection to the output devices whilst still delivering the rated power for higher impedances - sometimes without resorting to the usual current limiting circuits. Current limiting in one form or another is still a necessary protection even for DIY gear though few of us actually utilise the available power and are probably only listening to 1-5W average levels in domestic use. This is probably why we can simply disable it with impunity.
Early in the thread and on his website (see #311), Calvin emphasised the supply limits he found necessary and of course, +/-50V max is recommended by LJM. Whilst I have heard odd noises like faint hum and crackling when the voltage supply is too high for many audio circuits, it doesn't necessarily follow that this is the problem here.
As a more general comment, Japanese manufacturers often used relatively small power transformers with high rail voltages. These have poor regulation, so the supply sags under heavy, low impedance loads. This automatically gives some protection to the output devices whilst still delivering the rated power for higher impedances - sometimes without resorting to the usual current limiting circuits. Current limiting in one form or another is still a necessary protection even for DIY gear though few of us actually utilise the available power and are probably only listening to 1-5W average levels in domestic use. This is probably why we can simply disable it with impunity.
L12-2 idle noise
So today I brought the system up using a variac. At about 70-80 VAC (going by the variac dial only) the protection relays kicked in and the rails were about 31 VDC. The noise was still present from the left amp. Some other observations:
-the noise goes away when I disconnect the input to the left amp
-as mentioned earlier, the noise comes from the left amp no matter which input feed or output (speaker channel) I connect to the left amp
- a curious observation: if I quickly increase the variac, say by about 10VAC in < 1sec, the noise subsides but then returns after about 2-3 seconds. This also happens if I decrease the variac in a similar fashion. To me this says the problem is somehow related to the idle state, it seems like the AC transient (and presumably the corresponding rail transient) suppresses the noise but once things idle out the noise returns. The system still has the original Sony main caps, which could be related to this observation. But, since the right amp is very quiet in the Sony chassis, I'm skeptical that the caps are a major problem.
So, based on these observations and since it is difficult for me to take measurements on the L12 components with the amps installed in the Sony chassis, I am going to proceed with removing the amps and powering them side-by-side on the bench using the dual 30VDC supply. Perhaps something has been damaged on the noisy amp from running with less than 20% margin as you suggested.
Your remark about "original parts" is of course always a concern. I have several different TO-92s laying around that came from Mouser/DigiKey, so if it comes to that I'll see if a substitution is worth considering. Having the two amps side-by-side on the bench and taking some measurements of the output stages should help with this decision.
Regarding "relatively small transformers" I fully understand but the reason I have gone down this path is that the Sony transformer seems to be a pretty good sized toroid. Not to mention that in the original design the transformer also powered the Class A STK IC VAS which I have removed from the circuit,thus freeing up a little transformer capacity. There are no specs to be found for the transformer, but your remarks about regulation are duly noted.
Thank you for your continued correspondence and I will let you know what happens next.
So today I brought the system up using a variac. At about 70-80 VAC (going by the variac dial only) the protection relays kicked in and the rails were about 31 VDC. The noise was still present from the left amp. Some other observations:
-the noise goes away when I disconnect the input to the left amp
-as mentioned earlier, the noise comes from the left amp no matter which input feed or output (speaker channel) I connect to the left amp
- a curious observation: if I quickly increase the variac, say by about 10VAC in < 1sec, the noise subsides but then returns after about 2-3 seconds. This also happens if I decrease the variac in a similar fashion. To me this says the problem is somehow related to the idle state, it seems like the AC transient (and presumably the corresponding rail transient) suppresses the noise but once things idle out the noise returns. The system still has the original Sony main caps, which could be related to this observation. But, since the right amp is very quiet in the Sony chassis, I'm skeptical that the caps are a major problem.
So, based on these observations and since it is difficult for me to take measurements on the L12 components with the amps installed in the Sony chassis, I am going to proceed with removing the amps and powering them side-by-side on the bench using the dual 30VDC supply. Perhaps something has been damaged on the noisy amp from running with less than 20% margin as you suggested.
Your remark about "original parts" is of course always a concern. I have several different TO-92s laying around that came from Mouser/DigiKey, so if it comes to that I'll see if a substitution is worth considering. Having the two amps side-by-side on the bench and taking some measurements of the output stages should help with this decision.
Regarding "relatively small transformers" I fully understand but the reason I have gone down this path is that the Sony transformer seems to be a pretty good sized toroid. Not to mention that in the original design the transformer also powered the Class A STK IC VAS which I have removed from the circuit,thus freeing up a little transformer capacity. There are no specs to be found for the transformer, but your remarks about regulation are duly noted.
Thank you for your continued correspondence and I will let you know what happens next.
If noise ceases when you remove the left channel's input connection (I assume you mean both signal+ground connections), the noise could be due to a ground loop, though your description doesn't quite match with what I usually hear. Try disconnecting just the left channel's signal or signal ground alone. If you find the ground lead causes the problem, it likely is a loop, and should be correctable.
Ian et al:
I have spent considerable time studying the schematic(s) and component layout, and have taken some comparative measurements on my two boards.
For starters let me acknowledge that my boards appear to be some type of clone/knockoffs of the true LJM design. I say this based on comparison of my boards against some of the latest online images showing Version 3 of the "official" LJM board. Later if needed I will post more details about board differences including some photos. For now I don't think the differences between my boards and the "official" LJM V3 boards have any relation to the problem I am analyzing, so I'd like to report the findings of my measurements from the two boards i have here with the idle noise differences.
I ran the boards side-by-side on the bench with rails set at +/- 22.3 VDC. All transistors seem to have appropriate VBE and VCE voltage drops so there are no obvious catastrophic transistor failures. There are however a few DC voltage differences when comparing the two boards, discussed as follows.
Beginning at the input of the amp, the upper portion of the input circuitry, namely Q1 and Q3, have consistent voltages on both boards. The emitters of the differential pair, Q6 and Q8, also have consistent voltages across the two boards. The collectors of Q6 and Q8, which are also the emitters of Q15 and Q16 respectively, have different voltages. In turn the bases of Q15 and Q16 have different base voltages and this is where the first suspicious observation comes up. The bases of Q15 and Q16 are held at constant voltage below ground by Zener diode D3, 5V1, which is fed via R32 to VEE. On the good board the bases of Q15 and Q16 are -5.74 VDC and on the noisy board the bases are -5.29. To me these measurements are off by more than I would expect for 5V1 zener diodes, especially the -5.74 VDC measurement. Even more curious, the board in which the 5V1 zener is measuring -5.74 below ground, that is the board without the noise!
I think these 5V1 voltage differences across the two boards are then propagating downstream to the bias spreader and output devices. Without reporting all of the voltages here, I will say that on the good board the collectors of the 1186s and 2837s on the good board are 5mv and 4mv respectively, but on the noisy board they are -2mv and -4mv respectively.
I welcome opinions on this data. My inclination is to next replace one or both of 5V1 zeners in an attempt to bring the voltage measurements on each board into alignment, especially the voltages at the collectors of Q6, Q8, and the output devices. Since the 5V1 zener on the board with zero noise is actually running at about 5.7V, I'm thinking of increasing the zener voltage on both boards. I have some 5V6s on hand that I can use for this purpose.
Ian (and others lurking) please let me know what you think of this data and my planned next steps.
Finally, I have a question about R30 and C14 which connect the collectors of Q6 and Q8. On my boards it appears that C14 is not installed. R30 is connected to the collector of Q6 but it goes nowhere. From looking at photos of the version 3 boards it seems this part of the circuitry has changed. Can you shed any insights into what is going on here? Not only is this puzzling, it involves the diff pair so I am wondering if it could be related to the noise problem.
Thanks again for your help with this situation.
I have spent considerable time studying the schematic(s) and component layout, and have taken some comparative measurements on my two boards.
For starters let me acknowledge that my boards appear to be some type of clone/knockoffs of the true LJM design. I say this based on comparison of my boards against some of the latest online images showing Version 3 of the "official" LJM board. Later if needed I will post more details about board differences including some photos. For now I don't think the differences between my boards and the "official" LJM V3 boards have any relation to the problem I am analyzing, so I'd like to report the findings of my measurements from the two boards i have here with the idle noise differences.
I ran the boards side-by-side on the bench with rails set at +/- 22.3 VDC. All transistors seem to have appropriate VBE and VCE voltage drops so there are no obvious catastrophic transistor failures. There are however a few DC voltage differences when comparing the two boards, discussed as follows.
Beginning at the input of the amp, the upper portion of the input circuitry, namely Q1 and Q3, have consistent voltages on both boards. The emitters of the differential pair, Q6 and Q8, also have consistent voltages across the two boards. The collectors of Q6 and Q8, which are also the emitters of Q15 and Q16 respectively, have different voltages. In turn the bases of Q15 and Q16 have different base voltages and this is where the first suspicious observation comes up. The bases of Q15 and Q16 are held at constant voltage below ground by Zener diode D3, 5V1, which is fed via R32 to VEE. On the good board the bases of Q15 and Q16 are -5.74 VDC and on the noisy board the bases are -5.29. To me these measurements are off by more than I would expect for 5V1 zener diodes, especially the -5.74 VDC measurement. Even more curious, the board in which the 5V1 zener is measuring -5.74 below ground, that is the board without the noise!
I think these 5V1 voltage differences across the two boards are then propagating downstream to the bias spreader and output devices. Without reporting all of the voltages here, I will say that on the good board the collectors of the 1186s and 2837s on the good board are 5mv and 4mv respectively, but on the noisy board they are -2mv and -4mv respectively.
I welcome opinions on this data. My inclination is to next replace one or both of 5V1 zeners in an attempt to bring the voltage measurements on each board into alignment, especially the voltages at the collectors of Q6, Q8, and the output devices. Since the 5V1 zener on the board with zero noise is actually running at about 5.7V, I'm thinking of increasing the zener voltage on both boards. I have some 5V6s on hand that I can use for this purpose.
Ian (and others lurking) please let me know what you think of this data and my planned next steps.
Finally, I have a question about R30 and C14 which connect the collectors of Q6 and Q8. On my boards it appears that C14 is not installed. R30 is connected to the collector of Q6 but it goes nowhere. From looking at photos of the version 3 boards it seems this part of the circuitry has changed. Can you shed any insights into what is going on here? Not only is this puzzling, it involves the diff pair so I am wondering if it could be related to the noise problem.
Thanks again for your help with this situation.
I'm no expert on zener applications but the function here is just to bias the cascode transistors Q15/16 on sufficiently to conduct and add their C-E voltage drop to protect the 50V transistors in the LTP from the high potential between the supply rails. The precise zener voltage shouldn't be an issue, as long as the voltages across the cascodes are sufficient and stable at idle. However, a duff zener could easily introduce some excess noise and perhaps that's the culprit - easy enough to try a substitute as you plan to. On the other hand, +/-22V supplies may be too low to allow normal operation. If you can boost those power supplies by 5V or so, I'd check that out first.