Hi,
Just took a quick look at the power circuit. It seems the MC7805AC 5V regulator does not feed the 3 AMS1117-3.3 regulators directly. It feeds another IC (U8) and that connects to the 3.3V regs. I've not seen this before. Is it this ?
http://www.fairchildsemi.com/ds/FD/FDS4435.pdf
Pins 5-8 are all connected to the input of 3.3V regs and pins 1-3 is connected to the output of the MC7805. Pin 4 seems unconnected. So is it a buffer to improve the performance of the two stages of regulation ?
Just took a quick look at the power circuit. It seems the MC7805AC 5V regulator does not feed the 3 AMS1117-3.3 regulators directly. It feeds another IC (U8) and that connects to the 3.3V regs. I've not seen this before. Is it this ?
http://www.fairchildsemi.com/ds/FD/FDS4435.pdf
Pins 5-8 are all connected to the input of 3.3V regs and pins 1-3 is connected to the output of the MC7805. Pin 4 seems unconnected. So is it a buffer to improve the performance of the two stages of regulation ?
Last edited:
Hi,
I still can't find pin 4's connection but it has 5.6K R to gnd so that makes more sense, and there is a 2mV drop across the IC, so it seems it is a mosfet buffer. Nice. It should have a high-frequency roll off so it seems it does two jobs - buffers the two stages or regulation and filters noise. Can anyone with better circuit theory knowledge than me confirm this ?
I also checked the IR receiver IC and that takes 5V from the 7805. So I think the best way to improve the power is to remove/bypass the 7805 and feed a clean 5V into the QLS, and then upgrade/replace the AMS 3.3V regs. The mosfet buffer can stay.
I have an AMB sigma11 that uses an lm336 to set it at 5V so that should be an upgrade for the 7805.
EDIT Pin 4 connect to R32 way off in the far right corner and that has a potential divider to supply the gate.
I still can't find pin 4's connection but it has 5.6K R to gnd so that makes more sense, and there is a 2mV drop across the IC, so it seems it is a mosfet buffer. Nice. It should have a high-frequency roll off so it seems it does two jobs - buffers the two stages or regulation and filters noise. Can anyone with better circuit theory knowledge than me confirm this ?
I also checked the IR receiver IC and that takes 5V from the 7805. So I think the best way to improve the power is to remove/bypass the 7805 and feed a clean 5V into the QLS, and then upgrade/replace the AMS 3.3V regs. The mosfet buffer can stay.
I have an AMB sigma11 that uses an lm336 to set it at 5V so that should be an upgrade for the 7805.
EDIT Pin 4 connect to R32 way off in the far right corner and that has a potential divider to supply the gate.
Last edited:
Hi
I think using a local mini shunt regulator from Paul Hynes or the tridents from Twisted Pear (I'm using these) would be even more effective than working on the 7805/bypass, although that is probably worthwhile too.
Hi,
Thanks for the tips. I'm not convinced shunts are really that useful for constant loads like pure digital circuits, and I've also had a bad experience with Paul's gear so I'm not convinced they would be the best solution. However I will look into the Twisted Pear tridents. Given an accurate and stable voltage, I'm thinking low noise is the primary goal. What do you think ? Do Twisted Pear have any specs for their shunts ?
I removed the 7805, connected the pads with another ferrite, and added another Sanyo in the space left by the 7805, and I'm using an AMB sigma11 at the moment. Obviously it is better but I didn't notice any change in sq. I expect the ams regs are a bottleneck now.
I'm thinking about an sla or lifepo4 battery feeding a large cap bank for a low impedance, ripple free, low noise 12v source. And an ultra low noise voltage reference (like MAX6250 - it has 100dB ripple rejection, 1.5uV noise, etc) into the base of a BD139 via an r-c filter to create a 5V source with noise at or below uV levels. Then I'd replace the ams regs with potential dividers into a capacitance multiplier with bc550 or similar transistors. It might be interesting to add a Wenzel noise shunt too....but probably not enough space.
I think this should work well because line and load regulation aren't as critical as with stuff like audio op amps and really low noise is the goal. It isn't that expensive to do either.
Thanks for the tips. I'm not convinced shunts are really that useful for constant loads like pure digital circuits, and I've also had a bad experience with Paul's gear so I'm not convinced they would be the best solution. However I will look into the Twisted Pear tridents. Given an accurate and stable voltage, I'm thinking low noise is the primary goal. What do you think ? Do Twisted Pear have any specs for their shunts ?
I removed the 7805, connected the pads with another ferrite, and added another Sanyo in the space left by the 7805, and I'm using an AMB sigma11 at the moment. Obviously it is better but I didn't notice any change in sq. I expect the ams regs are a bottleneck now.
I'm thinking about an sla or lifepo4 battery feeding a large cap bank for a low impedance, ripple free, low noise 12v source. And an ultra low noise voltage reference (like MAX6250 - it has 100dB ripple rejection, 1.5uV noise, etc) into the base of a BD139 via an r-c filter to create a 5V source with noise at or below uV levels. Then I'd replace the ams regs with potential dividers into a capacitance multiplier with bc550 or similar transistors. It might be interesting to add a Wenzel noise shunt too....but probably not enough space.
I think this should work well because line and load regulation aren't as critical as with stuff like audio op amps and really low noise is the goal. It isn't that expensive to do either.
Last edited:
Hi,
I couldn't find any info re power demands so I've just done some measurements. Mine is using 163mA during start up, then 154mA while playing, and a 140-something idle. What was interesting is that there is some considerable noise generated during the start up that has a rising frequency, but this disappears as soon as the player is ready. Current demands during playback vary by less than 1mA, so it is a constant load.
I've tried a regulated sla battery (12V 5AH, with noise filters etc)- no difference to the sigma11 - so for sure the ams 3.3v regs are the limit.
I couldn't find any info re power demands so I've just done some measurements. Mine is using 163mA during start up, then 154mA while playing, and a 140-something idle. What was interesting is that there is some considerable noise generated during the start up that has a rising frequency, but this disappears as soon as the player is ready. Current demands during playback vary by less than 1mA, so it is a constant load.
I've tried a regulated sla battery (12V 5AH, with noise filters etc)- no difference to the sigma11 - so for sure the ams 3.3v regs are the limit.
With the screen removed, it draws 161, 151 and 139mA so the screen uses about 2 or 3mA. It's LCD with no backlight so that seems reasonable.
The CS8406 uses about 15mA iirc. I don't know about the remote receiver or the dsp IC but clearly they use the rest (less any consumed by the regs themselves) ~ 100mA
I took some time to measure the current fluctuation during playback and it is 0.6mA or about 0.5%, so load regulation is a non-issue.
The CS8406 uses about 15mA iirc. I don't know about the remote receiver or the dsp IC but clearly they use the rest (less any consumed by the regs themselves) ~ 100mA
I took some time to measure the current fluctuation during playback and it is 0.6mA or about 0.5%, so load regulation is a non-issue.
Last edited:
Here's an example of the replacement I've started using for the ams1117 3.3V reg (from Tent Labs site):
For DC, it is a potential divider to drop the voltage through the transistor. For AC, it is a filter to ground through the capacitor = a capacitance multiplier that helps to isolate it from the noise etc of other power drains.
Instead of a 7808 source, I have a Sigma11 5V, and I'm using a Bourns 10K divider, a BC550B and a 2.7K R load. Here's a picture of one I'm using on a Teralink X2 - it is a definite improvement - but it needs an excellent regulator before it to really impress.
For DC, it is a potential divider to drop the voltage through the transistor. For AC, it is a filter to ground through the capacitor = a capacitance multiplier that helps to isolate it from the noise etc of other power drains.
Instead of a 7808 source, I have a Sigma11 5V, and I'm using a Bourns 10K divider, a BC550B and a 2.7K R load. Here's a picture of one I'm using on a Teralink X2 - it is a definite improvement - but it needs an excellent regulator before it to really impress.
Last edited:
Hi,
I found a bigger 5V screen - the characters are almost twice the height (9mm). It is 16x1 but the datasheet says it is electrically connected as 8x2. It has a backlight too so I'm hoping this will be big enough to see from 2.5 metres away.
VARITRONIX|MDLS16168-LV|LCD MODULE, ALPHANUMERIC, 16X1 | e絡盟 Hong Kong
http://www.farnell.com/datasheets/26258.pdf
I've asked Clark if he thinks it is compatible (forum and email).
And for those into pics, here is the cs8406 reg replacement. As you can see, it isn't exactly pretty, and my skill with the iron leaves a lot to be desired, but it works very well nonetheless. I didn't use a load resistor this time as I set the trimmer before installing it, and then fine adjusted it to give exactly 3.3V at the CS8406. It doesn't add any regulation but it reduces noise and isolates the CS8406 from the other devices. It means it is easier to hear the difference between various 5V power sources too. The Bourns trimmer is just the right size to fit under the screen.
I found a bigger 5V screen - the characters are almost twice the height (9mm). It is 16x1 but the datasheet says it is electrically connected as 8x2. It has a backlight too so I'm hoping this will be big enough to see from 2.5 metres away.
VARITRONIX|MDLS16168-LV|LCD MODULE, ALPHANUMERIC, 16X1 | e絡盟 Hong Kong
http://www.farnell.com/datasheets/26258.pdf
I've asked Clark if he thinks it is compatible (forum and email).
And for those into pics, here is the cs8406 reg replacement. As you can see, it isn't exactly pretty, and my skill with the iron leaves a lot to be desired, but it works very well nonetheless. I didn't use a load resistor this time as I set the trimmer before installing it, and then fine adjusted it to give exactly 3.3V at the CS8406. It doesn't add any regulation but it reduces noise and isolates the CS8406 from the other devices. It means it is easier to hear the difference between various 5V power sources too. The Bourns trimmer is just the right size to fit under the screen.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Hi,
I heard from Clark and he is not sure if the screen will work because the datasheet doesn't contain firmware info, so I'll try and see.
I checked out the datasheet for the main IC :
http://ww1.microchip.com/downloads/en/DeviceDoc/70286C.pdf
and it seems it has an internal voltage regulator because it uses 2.5V not 3.3V. I lifted a leg on the ams reg that supplies it and found the current demand to be quite high and it also drops after reading the card, so it is not suitable for the potential divider/capacitance multiplier, or a noise shunt, and in facts needs a regulator.
So I found a substitute with about 8dB lower noise. Most of the other specs are largely irrelevant because of the excellent 5V regulator. It's a Ti TL1963a. It isn't pin for pin so the reg has to be "moved down" and the Vin can be through a ferrite (elevated a few mm above the board). A picture paints a thousand...
It is sounding a LOT better than standard - the harshness in the top end has been replaced with fine accurate detail and sound stage (mostly the new tcxo) and it is slightly warmer overall with more extended bass too (the cap multiplier for the CS8406 helped with this).
This sounds really very good overall now. And I'm still using spdif and the original pulse trans too !
I heard from Clark and he is not sure if the screen will work because the datasheet doesn't contain firmware info, so I'll try and see.
I checked out the datasheet for the main IC :
http://ww1.microchip.com/downloads/en/DeviceDoc/70286C.pdf
and it seems it has an internal voltage regulator because it uses 2.5V not 3.3V. I lifted a leg on the ams reg that supplies it and found the current demand to be quite high and it also drops after reading the card, so it is not suitable for the potential divider/capacitance multiplier, or a noise shunt, and in facts needs a regulator.
So I found a substitute with about 8dB lower noise. Most of the other specs are largely irrelevant because of the excellent 5V regulator. It's a Ti TL1963a. It isn't pin for pin so the reg has to be "moved down" and the Vin can be through a ferrite (elevated a few mm above the board). A picture paints a thousand...
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
It is sounding a LOT better than standard - the harshness in the top end has been replaced with fine accurate detail and sound stage (mostly the new tcxo) and it is slightly warmer overall with more extended bass too (the cap multiplier for the CS8406 helped with this).
This sounds really very good overall now. And I'm still using spdif and the original pulse trans too !
New screen works 
Hi,
So here's a pic for ya. Much bigger than the original eh ? I haven't connected up the backlight yet because I'd like to control the brightness, and I'm still working on the power etc, and now I'm thinking I'll need to re-case it too. So it's still very much a work in progress.
Anyway, no problems connecting it up - 14 pins all the same.
Hi,
So here's a pic for ya. Much bigger than the original eh ? I haven't connected up the backlight yet because I'd like to control the brightness, and I'm still working on the power etc, and now I'm thinking I'll need to re-case it too. So it's still very much a work in progress.
Anyway, no problems connecting it up - 14 pins all the same.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Great ? Hmm... bit of a bodge to be honest... but it's readable from across the room so that's all I wanted.
But now I'd like to be able to change the display readout, so it is folder>track>number>total
because at the moment it is
track>folder>number>total
so
folder>number
actually appears as one long number.
I've also found some awesome regs to replace the AMS1117 - Analog Devices ADP150 and ADP151 - just 9uV rms of noise ! 20dB lower than the 1117. That's lower than anything I've ever seen before, except regs with Wenzel noise shunts after them. Problem is they are too small for the board... but I'll oder some and see if I can get them on somehow. There is a LP reg from National that is similar noise levels but that is only micro size and impossible for diy.
I also found that the output of the trafo is tied to ground. Very weird thing to do ! So I replaced that with a Murata DA101 and kept the output side isolated from ground and changed the rca to bnc. That's made a noticeable difference too and seems a no-brainer upgrade for all.
But now I'd like to be able to change the display readout, so it is folder>track>number>total
because at the moment it is
track>folder>number>total
so
folder>number
actually appears as one long number.
I've also found some awesome regs to replace the AMS1117 - Analog Devices ADP150 and ADP151 - just 9uV rms of noise ! 20dB lower than the 1117. That's lower than anything I've ever seen before, except regs with Wenzel noise shunts after them. Problem is they are too small for the board... but I'll oder some and see if I can get them on somehow. There is a LP reg from National that is similar noise levels but that is only micro size and impossible for diy.
I also found that the output of the trafo is tied to ground. Very weird thing to do ! So I replaced that with a Murata DA101 and kept the output side isolated from ground and changed the rca to bnc. That's made a noticeable difference too and seems a no-brainer upgrade for all.
I have plans to build my own SDcard player one day. It won't be using the dsPIC as that's far too power hungry.
You wanna use low noise regs on the clock osc? Guillermo found some low noise regs and mentioned them on his blog - New Breed of Ultra Low Noise Regulators H i F i D U I N O. According to his table, the ADI parts are quite loud
You wanna use low noise regs on the clock osc? Guillermo found some low noise regs and mentioned them on his blog - New Breed of Ultra Low Noise Regulators H i F i D U I N O. According to his table, the ADI parts are quite loud
Hi,
Cool stuff ! Thanks ! He has the discrete LP regs I was looking at but rejected cos they're way too small, plus some other circuits like the AN which are also beyond beyond my skills. But there is one for diyers to check out the 723 looks really interesting as it is adjustable so I could use it with a Wenzel noise shunt to get down to about ~300nV !
The 723 looks like a real find - 2.5uV noise from a dip8 package ! Thank you !
Cool stuff ! Thanks ! He has the discrete LP regs I was looking at but rejected cos they're way too small, plus some other circuits like the AN which are also beyond beyond my skills. But there is one for diyers to check out the 723 looks really interesting as it is adjustable so I could use it with a Wenzel noise shunt to get down to about ~300nV !
The 723 looks like a real find - 2.5uV noise from a dip8 package ! Thank you !
"New Breed of Ultra Low Noise Regulators" - Haha, funny that that includes the grandpa of integrated regulators, the 723!
From that reg I was usually using only the reference voltage, the OpAmp part is crappy at best.
I seriosly doubt those measurements.
From that reg I was usually using only the reference voltage, the OpAmp part is crappy at best.
I seriosly doubt those measurements.
An externally hosted image should be here but it was not working when we last tested it.
Last edited: