Hi,
here a pic of my deq2496 with direct transformer output.
I directly solder wire from the ak4393 pins to the tranny with 120ohm in series.
I use a Tamura tpc-101 that I have on hand and connect 600:600.
I reference the cs4396 circuit in the jensen notes below.
http://www.jensen-transformers.com/as/as093.pdf
The sound of cause is a huge improvement from the original output.
here a pic of my deq2496 with direct transformer output.
I directly solder wire from the ak4393 pins to the tranny with 120ohm in series.
I use a Tamura tpc-101 that I have on hand and connect 600:600.
I reference the cs4396 circuit in the jensen notes below.
http://www.jensen-transformers.com/as/as093.pdf
The sound of cause is a huge improvement from the original output.
Attachments
Interesting project!
A couple of questions. When I looked at the AKM 4393 data sheet. It said the the unit needs to be loaded by at 1Kohm resistor and I noticed you only had a much smaller resistor in your circuit (as shown in the Jensen schematic). Did you play with this value at all?
Also the AKM sheet mentions a "built in" anti-aliasing filter on the chip. Is this why you and others can get by with only a single pole filter? Or are using the transformer is part of the low pass circuit?
thanks
A couple of questions. When I looked at the AKM 4393 data sheet. It said the the unit needs to be loaded by at 1Kohm resistor and I noticed you only had a much smaller resistor in your circuit (as shown in the Jensen schematic). Did you play with this value at all?
Also the AKM sheet mentions a "built in" anti-aliasing filter on the chip. Is this why you and others can get by with only a single pole filter? Or are using the transformer is part of the low pass circuit?
thanks
Can not find anything in the datasheet that says it must be loaded with 1kohm (or 600ohm). Only that it should not be loaded with less. But they do not recommend an optimal load either.
But is easy to adjust the loading with a transformer as the load on the secondary is reflected over to the primary with the addition of the DCRs.
This is an elegant solution also discussed in
http://www.diyaudio.com/forums/showthread.php?postid=1540153#post1540153
But is easy to adjust the loading with a transformer as the load on the secondary is reflected over to the primary with the addition of the DCRs.
This is an elegant solution also discussed in
http://www.diyaudio.com/forums/showthread.php?postid=1540153#post1540153
I am a little puzzled about the min. 1kohm DC spec. But I came to the conclusion that the serial resistors plus the RDC of the transformers primary should not be less than this value. I will go by that with my DCX, with Lundahl LL1544A connected as 600:600. In that case the serial resistors will be 430ohm as the RDC is 130ohm, totals 990ohm. Another option is to connect them as 10k:10k with 240ohm serial with the 520ohm RDC. The eventual problem then is that the load should be well above 100k.
If you connect the two phase outputs of the DAC to the ends of the primary, and leave the center tap either unconnected, or connected to the Vcom, the DC load is infinite. There cannot be any DC current between two points that are the same voltage, which means infinite resistance.
Jan Didden
Jan Didden
Jan,
Out of a theoretical perspective you are absolutely right
!
But when it comes to the AKM spec. sheet there has to be some guesswork done. Don´t think one should look at the AKM spec. your way. Instead I think they mean 1k w/o outputcap and 600ohm w/ capacitor. The specified min load is maybe between + and - output, but it could of course also be from + resp.- to ground. Also there is no info about the 3,5mA current capability.
Or maybe you have more in depth info from AKM? Then please share this information!
With 1k load to GND from each output, if we assume 2.5VDC at the output, there should be an IDC of 2,5mA. So with serial resistors and RDC summing up to 1k on each leg it should be possible to connect a transformer to ground. Although it probably isnt a good way to do it . Either no CT or CT to VCOM, together with careful offset adjustment must be the way to go.
Out of a theoretical perspective you are absolutely right
!But when it comes to the AKM spec. sheet there has to be some guesswork done. Don´t think one should look at the AKM spec. your way. Instead I think they mean 1k w/o outputcap and 600ohm w/ capacitor. The specified min load is maybe between + and - output, but it could of course also be from + resp.- to ground. Also there is no info about the 3,5mA current capability.
Or maybe you have more in depth info from AKM? Then please share this information!
With 1k load to GND from each output, if we assume 2.5VDC at the output, there should be an IDC of 2,5mA. So with serial resistors and RDC summing up to 1k on each leg it should be possible to connect a transformer to ground. Although it probably isnt a good way to do it . Either no CT or CT to VCOM, together with careful offset adjustment must be the way to go.
Jan,
Think I some wrong assumtions yesterday. Checked the filter schematics in the AK spec. sheet. They indicate the min. loading values are relative to ground. So if this is the case min. AC-loading for a floating load between + and - is 1,2k.
As a typical 600:600 transformer has an inductance of maybe 50H this corresponds to ca 6k at 20Hz. If we load the secondary with 10k we are still well above min. load.
Still one wonder what the optimum load is and if the optimum load includes some DC-loading?
Think I some wrong assumtions yesterday. Checked the filter schematics in the AK spec. sheet. They indicate the min. loading values are relative to ground. So if this is the case min. AC-loading for a floating load between + and - is 1,2k.
As a typical 600:600 transformer has an inductance of maybe 50H this corresponds to ca 6k at 20Hz. If we load the secondary with 10k we are still well above min. load.
Still one wonder what the optimum load is and if the optimum load includes some DC-loading?
revintage said:
Agree. Since the mid-point between the two outputs is a virtual ground, 1.2k between outputs translates to 600 per output.
revintage said:
I don't think there is any 'optimum' load. There is a minimum, as discussed above. Then as you increase the load, you increase the noise. How far do you go? It's one of those engineering trade-offs that vary from person to person. I'm in favour for lower than 10k, I'd probably use 2 or 3k. Why? A few * minimum for safety, but not too high for lowest noise. YMMV.
Jan Didden
About loading:
The normal secondary loading for a 600:600ohm transformer is 10k. If the following stage has higher Zin a secondary termination would be the best.
But to get a 2-3k loading(total or each leg?) according to Jans suggestion it is probably better to terminate the primary winding. This to get a more resistive(less frequency dependant) loading of the AKM chip.
Wouldn´t the low DCR of a transformer be of benefit for low noise, or is it the AC-loading that rules?
The normal secondary loading for a 600:600ohm transformer is 10k. If the following stage has higher Zin a secondary termination would be the best.
But to get a 2-3k loading(total or each leg?) according to Jans suggestion it is probably better to terminate the primary winding. This to get a more resistive(less frequency dependant) loading of the AKM chip.
Wouldn´t the low DCR of a transformer be of benefit for low noise, or is it the AC-loading that rules?
AFAIK the DCR is seen by the source (DAC) in series with whatever is at the secondary, so it plays a minor role in noise resistance.
Lars, what exactly means "600:600', apart from the fact that the turns ratio is 1:1? Doesn't that mean a transformer optimized in terms of iron and copper mass for 600 ohms source- and load impedance? A 10k:10k transformer would also have a 1:1 turns ratio but less iron (is smaller) because it is optimized for 10k impedances, meaning less current, meaning less magnetization, meaning smaller, cheaper.
So, you could just terminate a 600:600 with say, 2k, and the DAC would see 2k, without any issues, right? Or am I missing something?
Jan Didden
Lars, what exactly means "600:600', apart from the fact that the turns ratio is 1:1? Doesn't that mean a transformer optimized in terms of iron and copper mass for 600 ohms source- and load impedance? A 10k:10k transformer would also have a 1:1 turns ratio but less iron (is smaller) because it is optimized for 10k impedances, meaning less current, meaning less magnetization, meaning smaller, cheaper.
So, you could just terminate a 600:600 with say, 2k, and the DAC would see 2k, without any issues, right? Or am I missing something?
Jan Didden
Jan,
The DCR of the primary and secondary is in series with the overreflected secondary load if you look at it from a AC point of view but still for DC it is only the DCR. So you are right that for HF-noise it shouldn´t matter.
What you miss is that winding resistances and capasitances play a great role in how a transformer works. So your 2k primary to 2k seciondary (+DCR) is only valid in the midband, not at the extreme ends of the spectra.
"600:600" is just an popular expression and does not mean so much. It only indicates that it is to be driven from an lowZ source and has 1:1 voltageratio.
So it is maybe not adviseable to use impedance ratios. Better with voltage ratios. But then we have to know inductance as this has effect on loading at low frequencies. One should also know winding capacitances as this affects high frequency performance.
About the 600:600 against 10k:10k one can say that the DCRs normally are to high with the 10k:10k to be looked upon as sources for the preceding amp.
Example of a typical transformer to be used at the AK output:
Lundahl LL1544A, "600:600"-connected has a DCR of 130/side. with some serial resitance on the primary we maybe add up to 500-800ohm. Still it´s recommended load is 10k.
Connected "10k:10k" the DCRs are 520/side which adds up to 1,3-1,6k Zout.
My rule of thumb is to make loading 10x Zout.
So if we assume 10k in for the poweramp the loading is to high on the latter. Also there is no need for a highimpedance 1:1 as the AK is a low Zout device.
I am sure that you are aware of all this even if I understand you are not so fond of using transformers. As you avoid them I try to avoid caps and transistors. Simply a matter of taste and experience.
The DCR of the primary and secondary is in series with the overreflected secondary load if you look at it from a AC point of view but still for DC it is only the DCR. So you are right that for HF-noise it shouldn´t matter.
What you miss is that winding resistances and capasitances play a great role in how a transformer works. So your 2k primary to 2k seciondary (+DCR) is only valid in the midband, not at the extreme ends of the spectra.
"600:600" is just an popular expression and does not mean so much. It only indicates that it is to be driven from an lowZ source and has 1:1 voltageratio.
So it is maybe not adviseable to use impedance ratios. Better with voltage ratios. But then we have to know inductance as this has effect on loading at low frequencies. One should also know winding capacitances as this affects high frequency performance.
About the 600:600 against 10k:10k one can say that the DCRs normally are to high with the 10k:10k to be looked upon as sources for the preceding amp.
Example of a typical transformer to be used at the AK output:
Lundahl LL1544A, "600:600"-connected has a DCR of 130/side. with some serial resitance on the primary we maybe add up to 500-800ohm. Still it´s recommended load is 10k.
Connected "10k:10k" the DCRs are 520/side which adds up to 1,3-1,6k Zout.
My rule of thumb is to make loading 10x Zout.
So if we assume 10k in for the poweramp the loading is to high on the latter. Also there is no need for a highimpedance 1:1 as the AK is a low Zout device.
I am sure that you are aware of all this even if I understand you are not so fond of using transformers. As you avoid them I try to avoid caps and transistors. Simply a matter of taste and experience.
Lars,
Its not that I am 'not fond of transformers'; I have no particular opinion, use what seems best in a certain application and that depends indeed on your personal preference. No problem there I hope.
I'm really trying to grasp why a 600:600 with, say, 2k sec load wouldn't work well. Surely at higher freqs the transformer will more and more approach an 'ideal' xformer, 1:1 ratio, until the parasitic capacitances come into play. So maybe by loading the xformer with higher R lowers the available bandwidth at the high end?
The other question is at lf: there, because of the limited (not infinite) inductance, you will experience increased distortion and deviation from the 1:1 nominal value (output fall off). I think loading it with higher R extends the lf bandwidth towards lower freqs.
Anyway, as you say, the DCR from the DAC pov is immaterial as there is no DC current draw anyway. What would be your preferred way then, a '600:600' with 10k load?
Jan Didden
Its not that I am 'not fond of transformers'; I have no particular opinion, use what seems best in a certain application and that depends indeed on your personal preference. No problem there I hope.
I'm really trying to grasp why a 600:600 with, say, 2k sec load wouldn't work well. Surely at higher freqs the transformer will more and more approach an 'ideal' xformer, 1:1 ratio, until the parasitic capacitances come into play. So maybe by loading the xformer with higher R lowers the available bandwidth at the high end?
The other question is at lf: there, because of the limited (not infinite) inductance, you will experience increased distortion and deviation from the 1:1 nominal value (output fall off). I think loading it with higher R extends the lf bandwidth towards lower freqs.
Anyway, as you say, the DCR from the DAC pov is immaterial as there is no DC current draw anyway. What would be your preferred way then, a '600:600' with 10k load?
Jan Didden
Jan,
To give a picture of what will happen IRL I did a sim of six different cases.
We have:
Diff Zout of AK 100ohm
270ohms serial resistors on each leg
Data from LL1544A with hypothetical winding capacitances
AK loaded with ca 2,5kohm in each case.
With secondary loads higher than 2k the loading is on the primary
First digits are transformer version, 60010k means 600:600 with 10k secondary load.
So it should be obvious that 600:600 ohm with 10k or higher load is my choice.
To give a picture of what will happen IRL I did a sim of six different cases.
We have:
Diff Zout of AK 100ohm
270ohms serial resistors on each leg
Data from LL1544A with hypothetical winding capacitances
AK loaded with ca 2,5kohm in each case.
With secondary loads higher than 2k the loading is on the primary
First digits are transformer version, 60010k means 600:600 with 10k secondary load.
So it should be obvious that 600:600 ohm with 10k or higher load is my choice.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.