I decided to leave the 7810 regulators on the DAC boards alone, so I need 12 VDC. Otherwise the ultra cap boards might have been interesting.
I looked at the Lundahl 1694. I don't know if you connected the windings in series or in parallel. For my purposes, I'd have to use the parallel connection (40 mH), while the series connection (160 mH) would have too much DCR and would saturate the core. I calculated the filter function for the Lundahl + 10'000 uF and compared it with 15 mH Hammond 159ZG + 28'000 uF. The Lundahl will tend to ring, whereas the Hammond seems to behave a bit better.
Attachments
Hi Matthias,
Thanks for your contributions
on the oscillations and peak versus choke inductance. Point is in the simulation you also need to take the Load into account. With lower load there is higher oscillation... With high load (like 1A) it is dampened like in my graph.
Not sure how you arrived at your graphs? So I cannot judge what you did (what load for example?)
now the good news, as the power supply has a soft start anyway, this oscillation will not really show using the 1H for example, also at no load...
I tried and 10H has a bit hectic start but after 5~10 secs all is fine
Thanks for your contributions
on the oscillations and peak versus choke inductance. Point is in the simulation you also need to take the Load into account. With lower load there is higher oscillation... With high load (like 1A) it is dampened like in my graph.
Not sure how you arrived at your graphs? So I cannot judge what you did (what load for example?)
now the good news, as the power supply has a soft start anyway, this oscillation will not really show using the 1H for example, also at no load...
I tried and 10H has a bit hectic start but after 5~10 secs all is fine
I calculated my curves by looking at the AC voltages across the choke and the second capacitor in the CLC filter. The gain function of this LC part is the ratio of the AC ouput voltage across the C and the input voltage across L+C. The impedances of L and C are calculated following the good old schoolbook / Wikipedia formulae for L and C (and I added ESR for the cap, but that has virtually no effect at those low frequencies). I can post the GNU Octave (Matlab) code if anyone is interested.
I don't understand how the DC bias (=load current) would change the AC operation of the circuit. As long as the DC current does not saturate the core of the choke, the choke will work the same with and without a DC bias. The capacitor will not pass any DC anyway.
I agree that the low frequency ringing should not be "excited" at power on, but I still think the life of the regulator is better if it does not have pre-filter that tends to ring/oscillate if it's asked to do so.
I don't understand how the DC bias (=load current) would change the AC operation of the circuit. As long as the DC current does not saturate the core of the choke, the choke will work the same with and without a DC bias. The capacitor will not pass any DC anyway.
I agree that the low frequency ringing should not be "excited" at power on, but I still think the life of the regulator is better if it does not have pre-filter that tends to ring/oscillate if it's asked to do so.
That's why I use spice simulation 
Old school calculations in a regulated set up with many variables is very cumbersome. Now SPICE does it for me.
Trust me, the load has impact:
Circuit:
first
RC:
0,1 Ohm
4700uF
then LRC:
Inductor 1H
R 1 Ohm
C 4700
I ran 5 plots for 5,10,20,100, 1000 Ohm load:
Old school calculations in a regulated set up with many variables is very cumbersome. Now SPICE does it for me.Trust me, the load has impact:
Circuit:
first
RC:
0,1 Ohm
4700uF
then LRC:
Inductor 1H
R 1 Ohm
C 4700
I ran 5 plots for 5,10,20,100, 1000 Ohm load:
Attachments
Ok, now I understand why we get different results. Your schematic shows a load resistor in the output, probably used as a dummy load to get some current flow at the CLC output. This dummy resistor adds additional damping to the circuit.
There is no such dummy resistor in my analysis, so all the damping is left to the DCR of the choke.
The thing with SPICE is that it seems to stop my brain from thinking how a circuit really works. In this case, my brain is not sure which approach is better (with / without the dummy resistor). In the real-world application, your dummy resistor gets replaced with the PSU regulator. This means that the regulator would take resistors job to provide some extra damping to the CLC circuit.
However, I tend to believe that it would be better not to stress the regulator with this job.
P.S.: could you run your simulation with a CCS instead of a resistive load?
There is no such dummy resistor in my analysis, so all the damping is left to the DCR of the choke.
The thing with SPICE is that it seems to stop my brain from thinking how a circuit really works. In this case, my brain is not sure which approach is better (with / without the dummy resistor). In the real-world application, your dummy resistor gets replaced with the PSU regulator. This means that the regulator would take resistors job to provide some extra damping to the CLC circuit.
However, I tend to believe that it would be better not to stress the regulator with this job.
P.S.: could you run your simulation with a CCS instead of a resistive load?
Last edited:
Hello,
I have been doing a LOT of experiments with chokes in the DDDAC power supply and i asked Doede for some advice regularly.
As Doede tells us you must present a similar load to see what will happen in real life. Of course with the Tent shunts there will be a steady load ALL the time.
If you see something in the graphics that you dont like you can try adding an extra cap.
I think you should not worry about the dcr of the choke to much. Doede simulated a high dcr ( lundahl ll2771) for me. No problem and i am using it for choke input!
My friend uses the same one in a clc supply for 20 watt Hiraga power amp!
My favourite chokes are does made by Lundahl. But you must not exceed their current rating. Using close to limit no problem but dont go 20 % over. My friends Hiraga didnt like that. I think Dddac will present a more stable load than a class A power amp.
Greetings,Eduard
I have been doing a LOT of experiments with chokes in the DDDAC power supply and i asked Doede for some advice regularly.
As Doede tells us you must present a similar load to see what will happen in real life. Of course with the Tent shunts there will be a steady load ALL the time.
If you see something in the graphics that you dont like you can try adding an extra cap.
I think you should not worry about the dcr of the choke to much. Doede simulated a high dcr ( lundahl ll2771) for me. No problem and i am using it for choke input!
My friend uses the same one in a clc supply for 20 watt Hiraga power amp!
My favourite chokes are does made by Lundahl. But you must not exceed their current rating. Using close to limit no problem but dont go 20 % over. My friends Hiraga didnt like that. I think Dddac will present a more stable load than a class A power amp.
Greetings,Eduard
I am not so much worried about the DCR of the choke as such, or the choke make, or how a specific choke may be useful in a different environment. My aim is simply to design the CLC pre-filter for the regulator in the "Doede PSU" such that oscillation in the CLC is avoided.
Apart from the oscillation / damping requirements, the DCR must not be higher than about 0.9 Ohm in my case in order to make sure there is enough voltage drop across the PSU regulator.
Apart from the oscillation / damping requirements, the DCR must not be higher than about 0.9 Ohm in my case in order to make sure there is enough voltage drop across the PSU regulator.
Hello,
I am having VERY good results with using split bobbin transformers available at Mouser.
I have met several people who told me to use these in stead of the usual toriodal ones which seem to pass much more garbage from the '' power station '' into your sensible circuits.
Getting a choke with more mH and a bit more DCR combined with the right transformer will give lots of improvements. I think the cleaner the supply entering the original supply will be the better.
I have the idea that a choke will filter things that the original supply cant handle that well?
I dont know if there is pollution generated by an SMPS based supply connected somewhere will make life harder for a the standard regulation.
Maybe they are filtered by a choke or the regulation will create two different sounds?
Maybe you can crank up the voltage a bit by using SBYV28 diodes?? That will probably allow you to use LL1694 with two coils in series.
I honestly think that invest a little extra in a BIG choke plus the right transformer will give you more return on investment then 4 extra boards!!
Greetings, Eduard
I am having VERY good results with using split bobbin transformers available at Mouser.
I have met several people who told me to use these in stead of the usual toriodal ones which seem to pass much more garbage from the '' power station '' into your sensible circuits.
Getting a choke with more mH and a bit more DCR combined with the right transformer will give lots of improvements. I think the cleaner the supply entering the original supply will be the better.
I have the idea that a choke will filter things that the original supply cant handle that well?
I dont know if there is pollution generated by an SMPS based supply connected somewhere will make life harder for a the standard regulation.
Maybe they are filtered by a choke or the regulation will create two different sounds?
Maybe you can crank up the voltage a bit by using SBYV28 diodes?? That will probably allow you to use LL1694 with two coils in series.
I honestly think that invest a little extra in a BIG choke plus the right transformer will give you more return on investment then 4 extra boards!!
Greetings, Eduard
Hello,
I have used both Hammond and Triad.
Usually i made my choice depending on VA rating, circuit mounting or chassis mounting.
The chassis mounting types are easier to use you can solder the diodes directly to the terminals. Much easier to create airflow than circuit mounting.
I ALWAYS use choke input so then it will be easy As you can see in the picture short connection for a DC heater supply circuit ( Rod Coleman UK based) Here circuitboard Triad mounted onto the chassis which is ok but a bit hard to realize)
If you go to Mouser and search for split bobbin probably both Hammond and Triad will pop up.
Greetings, Eduard
P.s I have Triad made in Philippines and China. Hammond has plants in India and Mexico. I think these could be made EVERYWHERE if headquarters is willing to pay the right price they can get quality all over.
I have used both Hammond and Triad.
Usually i made my choice depending on VA rating, circuit mounting or chassis mounting.
The chassis mounting types are easier to use you can solder the diodes directly to the terminals. Much easier to create airflow than circuit mounting.
I ALWAYS use choke input so then it will be easy As you can see in the picture short connection for a DC heater supply circuit ( Rod Coleman UK based) Here circuitboard Triad mounted onto the chassis which is ok but a bit hard to realize)
If you go to Mouser and search for split bobbin probably both Hammond and Triad will pop up.
Greetings, Eduard
P.s I have Triad made in Philippines and China. Hammond has plants in India and Mexico. I think these could be made EVERYWHERE if headquarters is willing to pay the right price they can get quality all over.
Attachments
Hello,
Rod Coleman clearly tells us he likes split bobbin transformers
Dave Slagle from Intactaudio also uses the Rod Coleman circuits and made his own choke input circuit.
intact audio
Of course you can do something similar with Lundahl chokes!
Greetings, Eduard
Rod Coleman clearly tells us he likes split bobbin transformers
Dave Slagle from Intactaudio also uses the Rod Coleman circuits and made his own choke input circuit.
intact audio
Of course you can do something similar with Lundahl chokes!
Greetings, Eduard
Funny you bring up Dave Slagle. A few days ago, I realized that I don't need an active preamp anymore. I spent some time planing for a custom V.S.E RTP3D, so sad to see it go in favor of TVC/AVC... so now I am fully focused on that instead...
There is also Monolith Magnetics, amorphous & metglass cores, silver windings etc.... the limit is how deep your pocket is
There is also Monolith Magnetics, amorphous & metglass cores, silver windings etc.... the limit is how deep your pocket is
Finally after 4 years of building the DDDac 1794 I am mounting everything in a case.
My question is about the main power coming from the wall.
Normally the ground is tied to the case to protect the case from becoming hot in case a loose wire etc. Are any of the other grounds on the dac connected to this point. Like the 12volt or the 5 volt supplies.
My question is about the main power coming from the wall.
Normally the ground is tied to the case to protect the case from becoming hot in case a loose wire etc. Are any of the other grounds on the dac connected to this point. Like the 12volt or the 5 volt supplies.
Great moment ! Safety earth connected to the chassis is a good thing. There are many ways of doing things on grounding and many articles are handling this.
Personally I connect the ground of the power supply at the supply it self to the chassis. that is the only connection. Audio outputs are "floating"
The audio GND should have some reference to chassis / earth. However, I usually don't connect audio GND directly to chassis / earth to avoid ground loops. I'd suggest you read this: Earthing (Grounding) Your Hi-Fi - Tricks and Techniques
I connect the GND pin of the IEC connector directly to the chassis with a wire of at least 1.5mm2. Then I take a 10A 1000V bridge rectifier on which I connect a short plus and minus terminal. One AC terminal of the bridge goes to the chassis with a wire of at least 1.5mm2, the other AC terminal I connect to the minus pole of the power supply with a wire of at least 1.5mm2. I connect the alternating terminals with 100nF 275VAC X2 and 10ohm 2W MOX. The 10A bridge easily discharges the 16A home fuse without damage (tested). A stronger bridge can also be used, it's all cheap. For each power supply in DDDAC I use a separate bridge with a capacitor and a resistor. Means DDDAC power supply separately, WaveIO power supply separately and any other power supply if it exists separately. On the metal chassis, I form one point (star ground) where I ground each bridge. That point is not the same one where I connected the GND from the IEC connector. I’ve never heard any hum, and I’ve done a couple of DACs this way. I do amplifiers and preamplifiers similarly. The difference is that I form one central GND where I connect all the individual GNDs and then ground that point to the chassis via a 10A bridge, capacitor and resistor. IEC GND must be connected to a metal chassis for safety reasons, except in cases where the mains transformer has a particularly high insulation class. You can see that in various commercial devices that do not have a GND connection to the home installation, that is, they have a power cable with only two wires (L + N). Some even have an IEC connector without a GND pin.
