Checked the tweeter channel, and no changes there for better or worse..
But the bass channel was not satisifed with just one steel plate as can be seen from the first trace here, adding all four of them improved matters significantly and this is hardly audible even if I put my ear right next to the woofer cone.
Now, the next post will be interresting..
But the bass channel was not satisifed with just one steel plate as can be seen from the first trace here, adding all four of them improved matters significantly and this is hardly audible even if I put my ear right next to the woofer cone.
Now, the next post will be interresting..
Attachments
With 4 steel plates installed, I played arround a bit with apit of permalloy, and found that by sticking a couple of cm of the stuff in partially overlapping the edge of the steel plate stack, I could make the hum virtually disappear in to the noise-floor.
I allso tried this with the speaker connected, and the faint hum just disappeared, it was not audible any longer.
I tried inserting the permalloy between the chassis bulkhead and the stack of plates in the exactly same spot, no effect, I had to stick it inn between the cards and the stack...
Perhaps I can do something simmilar for the bass-channel..
I allso tried this with the speaker connected, and the faint hum just disappeared, it was not audible any longer.
I tried inserting the permalloy between the chassis bulkhead and the stack of plates in the exactly same spot, no effect, I had to stick it inn between the cards and the stack...
Perhaps I can do something simmilar for the bass-channel..
Attachments
Last edited:
very strange results Elbert! especially considering that all three of the sections are on the same pcb! The midrange getting worse with more sheets of steel is very counter-intuitive, but I'm sure there is a rational explanation for it!
Looks from the last post that you have some EMF leakage around the edge of your bulkhead?
Tony.
Looks from the last post that you have some EMF leakage around the edge of your bulkhead?
Tony.
I think there is a different mechanism that is inserting that pulse.
I think it is a wiring problem. That's why I thought looking at PSU output might help identify the source.
I think it is a wiring problem. That's why I thought looking at PSU output might help identify the source.
I agree, this was very counter intuitive.
But having said this, I believe this demonstrates how some experimentation is usefull in these situations, obviously stray-fields and leakages can be quite localized and affect very specific parts of a board or circuit!
So who knows, perhaps a little bit of mu-metal can straighten out the hum problem from your amplifier too?
BTW, did some very encouranging experiments with regards to the bass-channel.
Will post some proper measurements later this evening!
But having said this, I believe this demonstrates how some experimentation is usefull in these situations, obviously stray-fields and leakages can be quite localized and affect very specific parts of a board or circuit!
So who knows, perhaps a little bit of mu-metal can straighten out the hum problem from your amplifier too?
BTW, did some very encouranging experiments with regards to the bass-channel.
Will post some proper measurements later this evening!
Glade to hear that helpped!
How about some caps across the rectifier. I have seen many power supplys with caps on all 4 legs of the bridge rectifers, maybe we know why now. Andy
Hi Elbert, I had a bit of time to myself on Sunday, and rather than working on my regulated supply I decided to have a look at the PS of the chipamp again, and thought I'd share the results 🙂
The original power supply only had 1000uF per rail per channel. I'd always suspected that this was insufficient but have been using it happily for a year or more and never really noticed any deficiencies (apart from when doing the original power test where it didn't cope very well with a 20 Hz sinewave).
Before trying to butcher the PS I decided to add an additional 4700uF / rail to one channel of the PS, and make comparison measurements.
The increase to 5700uF per channel certainly reduced the hum on the output, so much so, I'd say that if I increased it more it would probably eliminate it (or at least to the point where it wasn't distinguishable from the PS noise in general.
First pic is the ripple on the supply with 1000uF / rail No load.
Second pic is the ripple on the supply with 5700uF/rail No load.
scope 100mv/div
Third pic is the noise on the output with 1000uF / rail.
Fourth pic is the noise on the other output with 5700uF / rail
scope 5mv/div
Both those are with input floating and no load on the amp.
with dummy load in the next post 🙂
The original power supply only had 1000uF per rail per channel. I'd always suspected that this was insufficient but have been using it happily for a year or more and never really noticed any deficiencies (apart from when doing the original power test where it didn't cope very well with a 20 Hz sinewave).
Before trying to butcher the PS I decided to add an additional 4700uF / rail to one channel of the PS, and make comparison measurements.
The increase to 5700uF per channel certainly reduced the hum on the output, so much so, I'd say that if I increased it more it would probably eliminate it (or at least to the point where it wasn't distinguishable from the PS noise in general.
First pic is the ripple on the supply with 1000uF / rail No load.
Second pic is the ripple on the supply with 5700uF/rail No load.
scope 100mv/div
Third pic is the noise on the output with 1000uF / rail.
Fourth pic is the noise on the other output with 5700uF / rail
scope 5mv/div
Both those are with input floating and no load on the amp.
with dummy load in the next post 🙂
Attachments
1st pic 1000uF / rail with 8 ohm dummy load on the output
2nd pic 5700uF / rail with 8 ohm dummy load on the output
scope 5mv/div What I thought was interesting about this was that the noise levels are greater, but also that the hum in the output seems more pronounced with the low capacitance supply. I suspect that there is a certain level of noise pickup from the dummy load, as the noise levels were MUCH higher with the amp turned off. probably about 1 1/2 divisions of noise just form the air. if I touch the probe it goes off the chart at even 20mv / division.
3rd pic amp with one channel of PS modified.
4th pic PS with 5700uF / rail per channel
As you can see I decided that it was worth it to change the other channel as well 🙂 I had the 4700uF caps spare anyway as they were surplus to my needs for my LM317 supply after I realised what a stupid mistake I'd made with it when AndrewT asked me a question about the caps 🙂 so they didn't go to waste!
The other good thing was that at 20Hz the power output was still around 33W into 8 ohms (about 35W at 100Hz and up) which for a 27V supply on an LM3886 is certainly acceptable I think 🙂 (it was only around 21W with the original 1000uF caps. Original (and new) power tests done with both channels driven.
edit: So my conclusion at this point is that whilst the amp is not 100% free from evidence of supply noise on the output, for a single transformer PS with separate bridges and caps for each channel (running off the same transformer secondaries) it doesn't do too badly 🙂
Tony.
2nd pic 5700uF / rail with 8 ohm dummy load on the output
scope 5mv/div What I thought was interesting about this was that the noise levels are greater, but also that the hum in the output seems more pronounced with the low capacitance supply. I suspect that there is a certain level of noise pickup from the dummy load, as the noise levels were MUCH higher with the amp turned off. probably about 1 1/2 divisions of noise just form the air. if I touch the probe it goes off the chart at even 20mv / division.
3rd pic amp with one channel of PS modified.
4th pic PS with 5700uF / rail per channel
As you can see I decided that it was worth it to change the other channel as well 🙂 I had the 4700uF caps spare anyway as they were surplus to my needs for my LM317 supply after I realised what a stupid mistake I'd made with it when AndrewT asked me a question about the caps 🙂 so they didn't go to waste!
The other good thing was that at 20Hz the power output was still around 33W into 8 ohms (about 35W at 100Hz and up) which for a 27V supply on an LM3886 is certainly acceptable I think 🙂 (it was only around 21W with the original 1000uF caps. Original (and new) power tests done with both channels driven.
edit: So my conclusion at this point is that whilst the amp is not 100% free from evidence of supply noise on the output, for a single transformer PS with separate bridges and caps for each channel (running off the same transformer secondaries) it doesn't do too badly 🙂
Tony.
Attachments
Last edited:
Thanks for that report on a thorough comparison of the effect of swapping between +-1mF and +-5m7F.
post267 pic 3 and post268 pic1, show a pulse above noise on the amp output. This looks like a charging pulse between transformer and smoothing capacitance via the rectifier. This should not appear on the output, nor should it appear on the supply rails of the amplifier. How is it getting to the output?
I wonder if input floating is allowing the rectifier loop to radiate air borne interference that the high impedance of an open input is able to pick up.
Does this "disappear" into the noise when the input is shorted and also when the input is terminated with 1k0?
I do wish others would at least try some of the suggestions that are posted rather than simply follow the "internet (dia)tribe".
As an aside.
I would never build a chipamp driving 8ohms with less than +-10mF/ch.
I prefer +-20mF/ch to suit the LF response I aim for (input filter set to ~90ms).
This preference/recommendation also works, for me, with discrete designs.
post267 pic 3 and post268 pic1, show a pulse above noise on the amp output. This looks like a charging pulse between transformer and smoothing capacitance via the rectifier. This should not appear on the output, nor should it appear on the supply rails of the amplifier. How is it getting to the output?
I wonder if input floating is allowing the rectifier loop to radiate air borne interference that the high impedance of an open input is able to pick up.
Does this "disappear" into the noise when the input is shorted and also when the input is terminated with 1k0?
I do wish others would at least try some of the suggestions that are posted rather than simply follow the "internet (dia)tribe".
As an aside.
I would never build a chipamp driving 8ohms with less than +-10mF/ch.
I prefer +-20mF/ch to suit the LF response I aim for (input filter set to ~90ms).
This preference/recommendation also works, for me, with discrete designs.
Last edited:
Thanks Andrew 🙂 The amp is actually powering a 4 ohm load (drops as low as 3 ohms) but my 100W dummy loads are 8 ohm and I only have two, so can't do 2 channel tests with a 4 ohm load.
The planned use for this amp (for a while anyway) will be for driving my MTM's crossed at 200Hz (once I finish (start?) the active crossover). It is currently driving them full range, but their F3 is around 90Hz. I have a hard time picking the difference between it and my 100W mosfet amp! Certainly today when I gave it a bit of a workout, It didn't seem lacking, but I don't ever claim to be able to pick before and after mod differences 🙂
The original purpose for this amp was as a speaker testing amp 🙂
Tony.
The planned use for this amp (for a while anyway) will be for driving my MTM's crossed at 200Hz (once I finish (start?) the active crossover). It is currently driving them full range, but their F3 is around 90Hz. I have a hard time picking the difference between it and my 100W mosfet amp! Certainly today when I gave it a bit of a workout, It didn't seem lacking, but I don't ever claim to be able to pick before and after mod differences 🙂
The original purpose for this amp was as a speaker testing amp 🙂
Tony.
the same amp component values for the >200Hz and for <200Hz drivers?
The Bass requirement certainly shows a benefit from high smoothing capacitance.
I would not expect the Mid/Treble to show any benefits from high smoothing capacitance and similarly the input filters of both upper and lower amps can be very different. This allows the internal amp component values (and types) to be selected to better suit the frequency range to be passed. The Mid/Treble may manage with no electrolytics in the amp's sections and very high quality low value electrolytics in the smoothing.
A treble only amp, may be built with no electrolytics whatsoever. A Peter Daniel implementation would just love to be the amp for Mid and/or Treble driver.
for which I reckon that smoothing capacitance must be doubled to maintain the same LF performance as driving 8ohm.
Last edited:
Hi Andrew, My 100W mosfet amp will be driving the bass units 🙂 The original 1000uF was Peter Daniel inspired, I was skeptical as to whether it would cut it, but it did pretty well all things considered. As I only have one chipamp amp I can't do A/B comparisons and I don't trust my memory enough to comment as to the effect on the bass of the cap upgrade, but the 100Hz hum on the output is certainly reduced, and although small in the first place it is a welcome addition. I probably should also have tested the ripple under load, as if 200mV of ripple on the PS (with no load) manifests as 100Hz noise on the output, then under load it will be worse.
Unfortunately this particular amp doesn't lend itself to reconfiguration, this picture might explain why 😉 But I can always take the extra capacitance out of the PS again.
Tony.
Unfortunately this particular amp doesn't lend itself to reconfiguration, this picture might explain why 😉 But I can always take the extra capacitance out of the PS again.
Tony.
Attachments
Hi had some thougths about that steel bolt my self, so thanks for prodding me in to investigating that further! 🙂
Regarding caps across the rectifier diodes, or "snubber caps", that was discussed somewhere else in this thread, and it seemed that to the extent it had any effect, it would not improve matters in my case as my issue seemed to be the charging current pulses rather than HF switching noise.
Tony,
Very interresting comparison you did there!
In addition to adding the extra capacitor, I would recommend that you try the "split capacitor" approach I implemented on my amplifier, the reduction in ripple was quite dramatic, probably the best value for money improvement I've come across considering the cost of adding a couple of low ohm power resistors! 🙂
Since last time, I've finished both my amps and put the lids on! 🙂
By adding some final bits of mu-metal here and there and tweaking the routing of the x-over-to amp signal wires, I obtained very good results.
Admittedly not perfect as I can still distinguish some very faint hum and noise (hiss) if I put my ear dead next to the speakers, but as a trade-off for keeping the x-over cards inside the chassii, thereby avoiding putting the x-over cards in a separate box with all the extra cabling etc..)I find it quite acceptable.
By adding some rubber-foam here and there I allso managed to reduce the mechanical hum from the big Toroids quite considerably.
My final challenge now is trying to balance the fan speed (noise) with appropriate cooling.
Very interresting comparison you did there!
In addition to adding the extra capacitor, I would recommend that you try the "split capacitor" approach I implemented on my amplifier, the reduction in ripple was quite dramatic, probably the best value for money improvement I've come across considering the cost of adding a couple of low ohm power resistors! 🙂
Since last time, I've finished both my amps and put the lids on! 🙂
By adding some final bits of mu-metal here and there and tweaking the routing of the x-over-to amp signal wires, I obtained very good results.
Admittedly not perfect as I can still distinguish some very faint hum and noise (hiss) if I put my ear dead next to the speakers, but as a trade-off for keeping the x-over cards inside the chassii, thereby avoiding putting the x-over cards in a separate box with all the extra cabling etc..)I find it quite acceptable.
By adding some rubber-foam here and there I allso managed to reduce the mechanical hum from the big Toroids quite considerably.
My final challenge now is trying to balance the fan speed (noise) with appropriate cooling.
Good news Elbert! 🙂 Interesting that you mention mechanical Hum from your torroids, this is probably an indicator of DC on your mains supply. I don't get any mecahnical hum from either of the torroids in my amps.
I'm certainly going to do crc (actually crcrc) in my regulated supply, And I might consider it for the chipamp, but it would require a bit more effort than what I did on the weekend 😉 without doing a new PS board I'd have to have the 1000uF on the rectifier side of the resistor, which some would say was better, but I'd like to have it the other way around. Will certainly think about it when doing the new PS for the mosfet amp.
Tony.
I'm certainly going to do crc (actually crcrc) in my regulated supply, And I might consider it for the chipamp, but it would require a bit more effort than what I did on the weekend 😉 without doing a new PS board I'd have to have the 1000uF on the rectifier side of the resistor, which some would say was better, but I'd like to have it the other way around. Will certainly think about it when doing the new PS for the mosfet amp.
Tony.
Hmmm.. or hummm.. Really hope I don't have DC on my mains.. but then again, we are talking about 600VA Toroids, and fairly hefty ones at that, so a minimum of mechanical hum is perhaps inevitable..
Hope CRC works out as well for you as it did for me, something I'll definitively implement in PSU's for any future projects! 🙂
See if I can post some nice pictures to round of my activity in this thread!
Hope CRC works out as well for you as it did for me, something I'll definitively implement in PSU's for any future projects! 🙂
See if I can post some nice pictures to round of my activity in this thread!
Could be, I've never used a torroid bigger than 300VA, have just read that DC on the mains will cause the transformer itself to hum.
Tony.
Tony.
I still dont like the power leads hooked up to the standoffs, just seems like trouble to me. Other than that vary nice. Enjoy. Andy
- Status
- Not open for further replies.