I was originally wanting to use a PGA2310 because of the better performance specs and the +-15V supplies but I was concerned there would be too much total gain through the preamp. It looks easy enough to add an SMT footprint to the board to accommodate all three of these devices as an option. I haven't seen any PGA2311 designs with a unity gain buffer on the input but what you are saying makes sense. I will look into adding one. Is it worth adding individual input attenuation as well? The +-5 or 15V supplies are an easy option by just adding a 0 ohm resistor to bypass the shunts. It's fed by the Jung regulators on the linestage which are +-15.
I'll look into dual footprints on the DAC buffer but that board is pretty tight with those big analogue traces already. A full SMT board would probably be an easier option. I can lay one of them out if anyone is interested after the through hole version is debugged.
An optional more advanced control board would likely be a fun project and a good excuse to learn some programming skills down the road too.
Thanks for the suggestions.
Last edited:
Did the reading ....
The 5532/4 is really a symasym/symasui with bjt inputs that runs on 15V. Not bad ,
even "golden earz" includes it as a valid DAC buffer option.
One thing I saw in a consensus of the "golden earz"
, they like to "roll op-amps"
(socketable - users choice) for the buffer -especially.
BTW , input buffer for the PGA - go for it ! Even as the DAC and most inputs are
usually low Z , some (or one) might not be. It would also be the first active semi after
the relays for any input other than the DAC , correct ?
If you made it socketed , it could be "sacrificial" at input instead of that hard
to replace SMD PGA , as well.
OS
The 5532/4 is really a symasym/symasui with bjt inputs that runs on 15V. Not bad ,
even "golden earz" includes it as a valid DAC buffer option.
One thing I saw in a consensus of the "golden earz"
, they like to "roll op-amps"(socketable - users choice) for the buffer -especially.
BTW , input buffer for the PGA - go for it ! Even as the DAC and most inputs are
usually low Z , some (or one) might not be. It would also be the first active semi after
the relays for any input other than the DAC , correct ?
If you made it socketed , it could be "sacrificial" at input instead of that hard
to replace SMD PGA , as well.
OS
When I built my second pre-amp, I used a ne5533, a 5534 dual in a 14pdip, now obsolete, it sounded better/different than my pioneer sx-950. Still today, 40 years later, it is a great part for the $.
the fun in it all, is to try to see for yourself and hear about peoples wacky audio description of what they are hearing. Like wine tasting, to me it is all gerd, i drink beer btw.
pga2310,11,20 are all the same as far as programming goes. going from memory, -96 to +15.5 dBv gain, in steps of 1dB. When i wrote the code, to test the part, i cylcled the counter at iirc +6dB. you will find that you do not need 15.5dB gain. this can be set in the program build. the dac is 2V full scale, so all that pga is doing is attenuation, that is why i ? even more gain in a buffer after the selector relays.
Also it is way to many steps, you want the s/w to have coarser steps or compressed in one/two regions, like a audio taper pot. I found about 32 steps is all you really need, and want, like the max9729 I am using in the portable.
input atten?
what source will cause an overload? with +/- 5V rails. the only thing i can think of is a phono stage clip due a bad lp.
Look at this design for an example,buffer,pga2320 iirc. read the write up to see how the ee made his design decisions.
K12 Preamp
See attached the pitchfork variation I am working on.
the mcu section is on the right with the BU9558 USB/SD codec. copied over from my portable.
The stuff on the left is the WM8805,PCM2902B,pga2320,pcm1794a. I have the routing/placement done, for so-8 that is, so you can follow my example. it is the section in the lower left.
cheers
rick
iirc, the pga2311 has the best thd specs followed by pga2320 lastly 2310, based on the graphs. pga2311 was the improved sub for cirrus earlier cs3310 iirc. i do not think you will hear the diff between them, but like "roll yer own opa", some may think or claim they hear a difference
the fun in it all, is to try to see for yourself and hear about peoples wacky audio description of what they are hearing. Like wine tasting, to me it is all gerd, i drink beer btw.pga2310,11,20 are all the same as far as programming goes. going from memory, -96 to +15.5 dBv gain, in steps of 1dB. When i wrote the code, to test the part, i cylcled the counter at iirc +6dB. you will find that you do not need 15.5dB gain. this can be set in the program build. the dac is 2V full scale, so all that pga is doing is attenuation, that is why i ? even more gain in a buffer after the selector relays.
Also it is way to many steps, you want the s/w to have coarser steps or compressed in one/two regions, like a audio taper pot. I found about 32 steps is all you really need, and want, like the max9729 I am using in the portable.
input atten?
what source will cause an overload? with +/- 5V rails. the only thing i can think of is a phono stage clip due a bad lp.
Look at this design for an example,buffer,pga2320 iirc. read the write up to see how the ee made his design decisions.
K12 Preamp
= a tube cathode follower perhaps
Sure, I do not mind sharing my design and code with diy group. I started with a sparkfun xmega100 bob.
See attached the pitchfork variation I am working on.
the mcu section is on the right with the BU9558 USB/SD codec. copied over from my portable.
The stuff on the left is the WM8805,PCM2902B,pga2320,pcm1794a. I have the routing/placement done, for so-8 that is, so you can follow my example. it is the section in the lower left.
cheers
rick
here's my 2 comments at 1 penny each
(sorry if i missed something from earlier discussion)
1. is LT1028 unity gain stable? (i.e. even suitable for buffer use this way)
2. LT1028 is an expensive, especially low noise for low input z part. maybe something else might be "better" here?
mlloyd1
Do you have any suggestions for a better choice of op amp? I just copied and pasted the input stage of the linestage and set the gain to unity. That gives me enough info to continue laying out boards until the analogue experts can show me the error of my ways.
i took a very quick look at the datasheet and see that although LT1028 is unity gain stable at -1 gain, it is not at +1, which is how it's configured in your circuit. but if folks really want to use an LT1?28 part, Linear does make the LT1128 that is. i did not look further to see what tradeoffs were made beyond the usual to make a part completely unity gain stable.
"better" is relative. i'll tell you what i'd try. just for a buffer, if i had to use an op amp, i'd try something not too exotic not too expensive and easy to find with a good track record like OPA132 or OPA134. if not limited to op amps i'd consider the discrete 2x njfet follower also (basically NP's B1 buffer). depends on what you may want to drive and if you want to brag about using fewer op amps or more discrete jfets.
however, knowing this crew, they'll try any opamp they can fit in the socket! it's all fun though.
mlloyd1
"better" is relative. i'll tell you what i'd try. just for a buffer, if i had to use an op amp, i'd try something not too exotic not too expensive and easy to find with a good track record like OPA132 or OPA134. if not limited to op amps i'd consider the discrete 2x njfet follower also (basically NP's B1 buffer). depends on what you may want to drive and if you want to brag about using fewer op amps or more discrete jfets.
however, knowing this crew, they'll try any opamp they can fit in the socket! it's all fun though.
mlloyd1
Thanks. I'm guilty of jamming anything with the correct pinout sometimes as well if I'm in a hurry to fire something up and see if it works. I'm slowly learning mare about audio analogue electronics.i took a very quick look at the datasheet and see that although LT1028 is unity gain stable at -1 gain, it is not at +1, which is how it's configured in your circuit. but if folks really want to use an LT1?28 part, Linear does make the LT1128 that is. i did not look further to see what tradeoffs were made beyond the usual to make a part completely unity gain stable.
"better" is relative. i'll tell you what i'd try. just for a buffer, if i had to use an op amp, i'd try something not too exotic not too expensive and easy to find with a good track record like OPA132 or OPA134. if not limited to op amps i'd consider the discrete 2x njfet follower also (basically NP's B1 buffer). depends on what you may want to drive and if you want to brag about using fewer op amps or more discrete jfets.
however, knowing this crew, they'll try any opamp they can fit in the socket! it's all fun though.
mlloyd1
Okay, lets try to avoid a hack'n wack design for your sake at least.
For the i/p buffer, my suggestion is to wire up using a single, have the comp cap option for say a ne5534a, wired in unity gain, dual so-8/dip8 footprint. Similar to the ne5534a dac i/v, as used for the pcm1794a. Look at the K9 pre-amp buffer U1A2A(L_Main,R_Main) for the way to wire up for optional gain or atten. If you want unity gain, use 0 ohm for the feed back R.
I am getting confused with your power supplies. Your naming conventions are confusing to me. Suggest to try something like +5VA,+5VA1,+5VD,+5VD1,-5VA,-5VA1,AGND, AGND1,A-DGND,DGND etc. using spaces makes it more confusing.
Also have you calculate the power for R23,26,29, it will be >1W! and what is the power for u6,7,8?
Since the PS voltages are going from one pcb to another, try to use the same names too.
On J1,3,4 add a 10 ohm Resistor as an option from pin 1 to gnd_analog (AGND)
That symbol for U9(x244) on the DAC/WM8805 pcb does not look correct, pins 17,13,15,11 named I6,I5,I6,I7? Should generate as symbol check error, duplicate pin name!
Also Leds D1,2,3,4 are back to back duals?, the symbol is not shown that way = confusing.
Well this is turning into design school but that is fine too as it is DIYaudio and we are all hear to learn and share.
for your sake at least.For the i/p buffer, my suggestion is to wire up using a single, have the comp cap option for say a ne5534a, wired in unity gain, dual so-8/dip8 footprint. Similar to the ne5534a dac i/v, as used for the pcm1794a. Look at the K9 pre-amp buffer U1A2A(L_Main,R_Main) for the way to wire up for optional gain or atten. If you want unity gain, use 0 ohm for the feed back R.
I think you may want to avoid discrete buffers unless your design is to feature an all discrete design. Since you are using an dip8 footprint, others can install discrete buffers if they so chose.
I am getting confused with your power supplies. Your naming conventions are confusing to me. Suggest to try something like +5VA,+5VA1,+5VD,+5VD1,-5VA,-5VA1,AGND, AGND1,A-DGND,DGND etc. using spaces makes it more confusing.
Also have you calculate the power for R23,26,29, it will be >1W! and what is the power for u6,7,8?
Since the PS voltages are going from one pcb to another, try to use the same names too.
On J1,3,4 add a 10 ohm Resistor as an option from pin 1 to gnd_analog (AGND)
That symbol for U9(x244) on the DAC/WM8805 pcb does not look correct, pins 17,13,15,11 named I6,I5,I6,I7? Should generate as symbol check error, duplicate pin name!
Also Leds D1,2,3,4 are back to back duals?, the symbol is not shown that way = confusing.
Well this is turning into design school but that is fine too as it is DIYaudio and we are all hear to learn and share.
If I didn't do hack'n wack designs I wouldn't get anything done. I'm a mechanic, not an electronics engineer.
I'm having issues with Diptrace and renaming power nets at the moment. It's joining nets randomly so I need to keep renaming them. I'm working with Diptrace tech right now to resolve this. The schematics are getting pretty tough to follow and component numbers are jumping all over the place. I'm going to update the schematics when I get the issues resolved and the design nailed down a little better.
R23 will be passing 1.5W/150mA max at the 100 ohm value. This is the maximum U6,7 & 8 can shunt as well. This is meant to be adjusted as needed. The DAC only needs 40mA and the PGA2311 only need 15mA. so these values will definitely need to be adjusted.
I plan to do as you suggest with the dual footprint op amps if possible for the buffer. OS has mentioned he would like the design to be through hole as much as possible so I'm using that wherever possible but if I can add an SMT footprint without making too much of a mess of the board I will. I'm trying to find the K9 preamp buffer you are referring to but am coming up blank. Do you have a link to it?
Good catch with the U9 thing. That's a preloaded Diptrace symbol as well. It hasn't raised any flags with the DRC yet.
I'll add those optional resistors to the analogue input grounds as well. In a spot like that I'll cheat and use 1206 SMT though.
I'm having issues with Diptrace and renaming power nets at the moment. It's joining nets randomly so I need to keep renaming them. I'm working with Diptrace tech right now to resolve this. The schematics are getting pretty tough to follow and component numbers are jumping all over the place. I'm going to update the schematics when I get the issues resolved and the design nailed down a little better.
R23 will be passing 1.5W/150mA max at the 100 ohm value. This is the maximum U6,7 & 8 can shunt as well. This is meant to be adjusted as needed. The DAC only needs 40mA and the PGA2311 only need 15mA. so these values will definitely need to be adjusted.
I plan to do as you suggest with the dual footprint op amps if possible for the buffer. OS has mentioned he would like the design to be through hole as much as possible so I'm using that wherever possible but if I can add an SMT footprint without making too much of a mess of the board I will. I'm trying to find the K9 preamp buffer you are referring to but am coming up blank. Do you have a link to it?
Good catch with the U9 thing. That's a preloaded Diptrace symbol as well. It hasn't raised any flags with the DRC yet.
I'll add those optional resistors to the analogue input grounds as well. In a spot like that I'll cheat and use 1206 SMT though.
Last edited:
Sorry to hear about diptrace issues. I have no idea how a cad tool can change your net names on you = eek.
Metz(K12 pre-amp) schematic
http://www.a-and-t-labs.com/K12_Preamp/Schematics_04232010.pdf
While you are on the phone with support ask them why that symbol does not flag an duplicate pin name error in the symbol editor. Probably because it does not care to check these things, as long as the pin numbers are not duplicated = poor tool set checking.
Metz(K12 pre-amp) schematic
http://www.a-and-t-labs.com/K12_Preamp/Schematics_04232010.pdf
I myself, do not like this approach and would change to a series pass reg such as used for the +/-15VA or LM317/337T (or equiv) or have the +/- 5VA on the PS pcb instead.
The shunts are regulating the Jung outputs. I think OS would shoot me if I added noisy old LM317/337 to them.
I can set more DRC rules to Diptrace but at the end of the day if the correct pin is connected to the right circuit, I really don't care if the pin names are wrong in the schematic. Most of the devices preloaded in Diptrace are obsolete. I'm going to take some time and create my own libraries soon.
Does this look better? I'm not going to have enough space on the Analogue IO board for this so it's going to need to be on another board which leaves me space on the IO board for LED arrays to indicate relay switching power.
Is my relay on the PGA input going to cause issues? I think the output relay opens momentarily while input relays are switching so there shouldn't be a huge pop through the amplifiers.
Is my relay on the PGA input going to cause issues? I think the output relay opens momentarily while input relays are switching so there shouldn't be a huge pop through the amplifiers.
OS,
Could you take a minute when you have the time and let me know what you think of some of the stock layout of the ESS dac layout and notes?
http://www.esstech.com/PDF/Application_Note_Component_Selection_and_PCB_Layout.pdf
I finally put the diptrace on a disk and will get it to you, both 32 and 64 bot. I'll include a surprise for you that you may like!
Could you take a minute when you have the time and let me know what you think of some of the stock layout of the ESS dac layout and notes?
http://www.esstech.com/PDF/Application_Note_Component_Selection_and_PCB_Layout.pdf
I finally put the diptrace on a disk and will get it to you, both 32 and 64 bot. I'll include a surprise for you that you may like!
PDF quality sucks. It is an evaluation PCB , longer than needed traces.
A bigger "footprint".
You like the ESS ? I know it's 32- bit ... so what ? It all comes down to
the output buffer - ha ha , they even have the lowly 5532 as a choice.
All that 32--bit goodness getting puked out of a 20 year old IC !
OS
OS,
I have no preference of one DSP chip over another, I don't know enough to make a comparison or choice. I was just wondering about the layout and if it was useful. I need to start somewhere and trying to find something that will work as a crossover and eq with possibilities for other functions later. Trying to read through some of the other threads with dsp's just isn't so easy to do, some are just going on forever and no final designs, or they end up being a commercial thread. I'm not looking to do a 8 channel solution for home theater. What you and JW are doing here seems to be much more complex than what I need just for the limited functionality I am trying to do. I'm not looking to add something like a small processor like an Arduino. Besides the fact that is starting to look so big, needing its own enclosure and all. I'm trying to work small and not have a rack mounted system or a stack of enclosures.
I have no preference of one DSP chip over another, I don't know enough to make a comparison or choice. I was just wondering about the layout and if it was useful. I need to start somewhere and trying to find something that will work as a crossover and eq with possibilities for other functions later. Trying to read through some of the other threads with dsp's just isn't so easy to do, some are just going on forever and no final designs, or they end up being a commercial thread. I'm not looking to do a 8 channel solution for home theater. What you and JW are doing here seems to be much more complex than what I need just for the limited functionality I am trying to do. I'm not looking to add something like a small processor like an Arduino. Besides the fact that is starting to look so big, needing its own enclosure and all. I'm trying to work small and not have a rack mounted system or a stack of enclosures.
I'm trying to design this so it can be as large or as small as you want it to be and to get a working design operating as a test platform /learning experience. After the first one I've usually got the bugs figured out and it's easy from there on. As we've talked about before, your application should be a custom specific build. I'd be happy to help you out with that after I get a working design figured out. It's a matter of cutting and pasting design blocks for the most part after I get an initial design done, then tweaking it for your application.
JW,
I really appreciate that. I really thought you had a specific application you were working on, not a learning experience. I wish I had even a bit of your knowledge about what you are doing. Give me a schematic and I can follow that and build it, but design it no way. I've been reading as much as I can. trying to figure out which dsp to use and the dacs needed. I thought that others would have designed something like what I am looking for but I have yet to find something not so overly complex to use, it is a bit frustrating to try and figure this all out. I also understand it is going to require some programming or finding a program that does what I want and programs the dsp. I'm a bit out of my league on this stuff. Give me a mechanical design project and I can run with it, this is really learning to read and write and interpret a different language.
I really appreciate that. I really thought you had a specific application you were working on, not a learning experience. I wish I had even a bit of your knowledge about what you are doing. Give me a schematic and I can follow that and build it, but design it no way. I've been reading as much as I can. trying to figure out which dsp to use and the dacs needed. I thought that others would have designed something like what I am looking for but I have yet to find something not so overly complex to use, it is a bit frustrating to try and figure this all out. I also understand it is going to require some programming or finding a program that does what I want and programs the dsp. I'm a bit out of my league on this stuff. Give me a mechanical design project and I can run with it, this is really learning to read and write and interpret a different language.
I haven't checked out Runescape too closely yet but if it'sanything like Sonos that would be much better than a bunch of different connections. Let it manage it all and do all the conversion. One simple network port and you're done.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.