LED reference (according to some) has lower noise than bandgap reference. i suppose it depends on the particular LED as well (I read one post with noise test results of various LED's forward biased at various current levels, but I can't find the post now). Long term drift and tempco are not issues for this application.
I suppose this will be relative easy to settle in our case - do a prototype with options for LED or TL431 and test the resulting performance
Alex
Better:
I have a Quan-tech Zener analyzer. I'll check some and see what I learn. Here is an interesting introduction to the issues:Re: Pete Lefferts LED current source
On a precision oscillator, especially one with a voltage trim, supply variations translate into frequency shifts so absolute voltage can be important.
On the series regulator with the 470 uF cap the noise is the junction of the single transistor, which is its strength. A 1K/470 uF filter will move the noise of the external reference down a few dB (over 60 at 1 KHz). This makes reference noise less of an issue. Doesn't help the transient response much however.
I have a Quan-tech Zener analyzer. I'll check some and see what I learn. Here is an interesting introduction to the issues:Re: Pete Lefferts LED current source
On a precision oscillator, especially one with a voltage trim, supply variations translate into frequency shifts so absolute voltage can be important.
On the series regulator with the 470 uF cap the noise is the junction of the single transistor, which is its strength. A 1K/470 uF filter will move the noise of the external reference down a few dB (over 60 at 1 KHz). This makes reference noise less of an issue. Doesn't help the transient response much however.
Last edited:
Here's a wild shot in the dark: If LED references suffer from optical sensitivity, then how feasible would it be to use the LED inside an optoisolator as the reference? Optoisolator are necessarily protected against external light input, so it would seem that this feature would effectively eliminate any optical sensitivity. They might even present less variance than a random LED. The obvious downside is that optoisolators are larger and more expensive than a discrete LED.
I would like you see the circuit simulation sent from me.
You can see an improvement in impedance (30 mOh. to 150 mOh) that can make alternative circuit interesting.
It also means less noise simulated(0.2 to 0.85 pV/Hz1/2)
The noise rejection is lower because it takes a reference to the entrance but not much less.
You do not like LED reference, Salas supply is the lowest noise power I have measured and is regulated by LED. LT431 is a OPA and they have Possible instability. I see in the simulations unstable between 10 and 50 Hz depending on the capacitor between the cathode and reference. This Possible instability is indicated in the graph 15 in the datasheed.
Now I have designed one PCB and I will design one another.
I can prove that they work.
Greetings.
You can see an improvement in impedance (30 mOh. to 150 mOh) that can make alternative circuit interesting.
An externally hosted image should be here but it was not working when we last tested it.
It also means less noise simulated(0.2 to 0.85 pV/Hz1/2)
An externally hosted image should be here but it was not working when we last tested it.
The noise rejection is lower because it takes a reference to the entrance but not much less.
An externally hosted image should be here but it was not working when we last tested it.
You do not like LED reference, Salas supply is the lowest noise power I have measured and is regulated by LED. LT431 is a OPA and they have Possible instability. I see in the simulations unstable between 10 and 50 Hz depending on the capacitor between the cathode and reference. This Possible instability is indicated in the graph 15 in the datasheed.
Now I have designed one PCB and I will design one another.
I can prove that they work.
Greetings.
Last edited:
Oneoclock:
Be careful of simulations. They can give you answers that don't exist in the real world.
You are passing a lot more current through Q5 than I am through Q1. This can actually lead to higher noise. You need to study the data sheet.
Here is a short description of the design and background. The noise at the output will never be less than the B-E noise of Q1 (Q5) and that won't be less than about .6 nV/rtHz. You also need to be sure that Q1 doesn't saturate or the noise will be much higher and the base current will go up.
The RC network to the base of Q1 is to attenuate the external noise and make the voltage at the base of Q1 essentially noiseless. The TL431 is used as a "servo' to correct for the secondary voltage "disturbers"; load and temperature. When I tried it without that feedback, between the transistor to transistor variations and the load/temperature effects, I was getting as much as 20% variation in the output voltage. Remember, at 3.3V .3V is 10%. With 20 dB of attenuation the noise contribution to the output at 10 Hz is small. At 1 KHz is non-existent.
I have built a number of these so far and with different loads have seen no evidence of instability.
Measuring the output noise is a real challenge. The lowest noise preamp circuit without a transformer will have an input noise in the .5 nV to .8 nV range. That plus this circuit leave a lowest possible reading of 1 nV and a lot of uncertainty. I used a transformer coupled preamp to lower that internal noise floor but still have problems with hum pickup I have not been able to resolve. All that will be examined after CES.
Be careful of simulations. They can give you answers that don't exist in the real world.
You are passing a lot more current through Q5 than I am through Q1. This can actually lead to higher noise. You need to study the data sheet.
Here is a short description of the design and background. The noise at the output will never be less than the B-E noise of Q1 (Q5) and that won't be less than about .6 nV/rtHz. You also need to be sure that Q1 doesn't saturate or the noise will be much higher and the base current will go up.
The RC network to the base of Q1 is to attenuate the external noise and make the voltage at the base of Q1 essentially noiseless. The TL431 is used as a "servo' to correct for the secondary voltage "disturbers"; load and temperature. When I tried it without that feedback, between the transistor to transistor variations and the load/temperature effects, I was getting as much as 20% variation in the output voltage. Remember, at 3.3V .3V is 10%. With 20 dB of attenuation the noise contribution to the output at 10 Hz is small. At 1 KHz is non-existent.
I have built a number of these so far and with different loads have seen no evidence of instability.
Measuring the output noise is a real challenge. The lowest noise preamp circuit without a transformer will have an input noise in the .5 nV to .8 nV range. That plus this circuit leave a lowest possible reading of 1 nV and a lot of uncertainty. I used a transformer coupled preamp to lower that internal noise floor but still have problems with hum pickup I have not been able to resolve. All that will be examined after CES.
Because I don't like to get my hands dirty
Seriously, though, I do prefer to design with off-the-shelf parts that a standard assembly house can stuff on a board for me. I'm quite handy with a soldering iron, when I have to be, but I prefer to let others handle the noxious fumes and stick to enjoying the final results.
Guys!
I'm happy to inform you that I brought the AB-11 back to life! Atmel support suggested that I should try updating firmware again, only this time without 'erase f' parameter. Guess what? It worked - leds glow, music plays
Cheers,
Marcin
Hi Marcingps
I´m in an similar situation as you.
The program say:Can´t open input file "*.elf"
May I ask you some clarification.
What to delete from the batchisp commando line?
"erase f" or "-operation erase f", or what exactly?
Thanks in advance.
Juan, did you check out this file package:
Add_to_flip345_bin.zip - sdr-widget - Programming kit on Win7 - Audio and Control Interface for Amateur Radio SDR and Audiophile USB-DAC - Google Project Hosting
It doesn't have the very latest in terms of compiled firmware (I'll update it from time to time), but it does have dos .bat files which are easy to modify.
What it does have is prog_noerase.bat which I wrote for your kind of case. You can try this file with .elf file as parameter, and you can try prog.bat with no parameter.
I have heard of 2-3 cases of this, one kit I have kept for later debugging. I can't see anything obviously wrong with it, though. And I'm able to program it some times.
Børge
Add_to_flip345_bin.zip - sdr-widget - Programming kit on Win7 - Audio and Control Interface for Amateur Radio SDR and Audiophile USB-DAC - Google Project Hosting
It doesn't have the very latest in terms of compiled firmware (I'll update it from time to time), but it does have dos .bat files which are easy to modify.
What it does have is prog_noerase.bat which I wrote for your kind of case. You can try this file with .elf file as parameter, and you can try prog.bat with no parameter.
I have heard of 2-3 cases of this, one kit I have kept for later debugging. I can't see anything obviously wrong with it, though. And I'm able to program it some times.
Børge
To any of you interested in the subjective sound quality assessments of the Widget, I've now added a pair of 470uF 16V SEPC caps for C21/C31. I have also added a 100uF Elna Cerafine to C15 position. This is still using my external (but pretty basic) 5V PSU (MUR860 diodes/1000uF Black Gate/LM7805/1000uF Black Gate/output).
Initial impressions are of the same super-refined & revealing sound, but with a touch extra bass heft and body. I guess it sounds more 'confident'. To be honest it's not a 'smack you over the head' difference, which proves the basic PSU design of this device without any mods is surprisingly good considering it's powered via USB as stock.
Cheers,
- John
Initial impressions are of the same super-refined & revealing sound, but with a touch extra bass heft and body. I guess it sounds more 'confident'. To be honest it's not a 'smack you over the head' difference, which proves the basic PSU design of this device without any mods is surprisingly good considering it's powered via USB as stock.
Cheers,
- John
Hi Rüdiger,
there is no Windows driver which officially supports the Audio Widget. (But a couple hacks have been reported..)
You can find the drivers from the SDR Widget Wiki page. Just follow my signature to click your way there. But the drivers are not needed for UAC1<->2 swap. That you can do by holding the Prog button until the light changes, keep holding it until the light goes out, then release, then reset.
Børge
there is no Windows driver which officially supports the Audio Widget. (But a couple hacks have been reported..)
You can find the drivers from the SDR Widget Wiki page. Just follow my signature to click your way there. But the drivers are not needed for UAC1<->2 swap. That you can do by holding the Prog button until the light changes, keep holding it until the light goes out, then release, then reset.
Børge
€59 Windows UAC2 driver
Has anybody tried this:
USB Audio - the low latency experience
We can try it as hassle-ware without paying. I won't have time for another week and a half due to travels.
Børge
Has anybody tried this:
USB Audio - the low latency experience
We can try it as hassle-ware without paying. I won't have time for another week and a half due to travels.
Børge
Web updates
This thread is getting huge! And we like it. I've done a quick check for some highlights and appended them to the FAQ. Just follow my signature and scroll to the bottom.
Nobody should feel left out! If your thought-through post isn't on the highlights lits, let me know!
Plus, www.qnktc.com finally makes a little bit of sense, too.
Børge
This thread is getting huge! And we like it. I've done a quick check for some highlights and appended them to the FAQ. Just follow my signature and scroll to the bottom.
Nobody should feel left out! If your thought-through post isn't on the highlights lits, let me know!
Plus, www.qnktc.com finally makes a little bit of sense, too.
Børge
Last edited:
Right you are, and it works. Very fine!
Now to some serious listening. See you!
Rüdiger