Pushing the limits of TDA1543 NOS DAC

[marcus1]

Will do, thanks.

Originally Posted by uncle_leon

Technically, you could just bypass the SN75179, but it would still be powered and so might have some minor influence on the sound via power supply or inductive/capacitative coupling. It's best to remove it altogether. It's actually a very easy job - since you are getting rid of it anyway, you can just cut its legs off or overheat the hell out of it.

[uncle_leon]

I have a question to Peter (or anyone else who feels confident on the subject).

I am contemplating reducing the input impedance of my amp, so I can take advantage of some of my new resistor discoveries that are only applicable at lower resistance values. At the moment, I use 10K. So here is the question:
How low can I take my amplifier input impedance before I overload the TDA1543 chip, or run into some other problems?

Normally, I would look at the max rated current in the datasheet and calculate the required impedance, but firstly, in this DAC, the chip is running outside its specification; and secondly, I admit I do not thoroughly understand how the I/V conversion works, and how it might affect rated current... So I decided it is best to ask the elders

[Peter Daniel]

Here's the link explaining how to calculate I/V resistors: Non Oversampling DAC-complementing CD-PRO

Your best bet is most likely trying few different lower resistor values in the amp first to see how it works, and then making your own.

[uncle_leon]

Thanks Peter. I am going to test different values, I was just hoping for a suggestion of a reasonable starting point.

How does an overloaded DAC behave, by the way? Can I possibly damage it, or will it just distort?

[Peter Daniel]

Never really heard it overloaded and it's almost impossible to damage it.

[Hierfi]

Originally Posted by uncle_leon

How does an overloaded DAC behave, by the way? Can I possibly damage it, or will it just distort?

The TDA1543 is a current output device and hence would not normally distort with reduced load resistance, rather the linearity could improve. This is because loading reduces the output voltage swing as the resistance seen by the TDA1543 reduces, causing it to operate in a more linear region. In other words, the effect of reduced resistance is only a reduction in amplitude. The current swing of the TDA1543 does not change.

Perhaps a more critical problem with loading is that the low frequency response becomes limited by the RC time constant of the output capacitance (the capacitor on the output of the DAC) in combination with the load resistor you choose on the output. In testing I would focus on any loss of low frequency extension, loss of detail or veiling. This could be countered by increasing the size of the coupling capacitor.

[uncle_leon]

Thanks Peter, that's what I wanted to hear!

@Hierfi
By reduction of amplitude you mean lower volume, or lower dynamic range? The former should not be a problem, but the latter I want to avoid at all cost.

As to RC filtering, I intended to counter that by swapping the output caps for a pair of 47uF/6.3V Black Gate Nx. Am I correct in my calculation that for a TDA1543 powered with 8V power supply, maximum voltage seen by the output capacitor will be 5V (i.e. the voltage swing component: 8V - 1.8V - 1.2V = 5V)? I would hate to blow such rare caps by overvolting them.

[Hierfi]

Originally Posted by uncle_leon

Thanks Peter, that's what I wanted to hear!

@Hierfi
By reduction of amplitude you mean lower volume, or lower dynamic range? The former should not be a problem, but the latter I want to avoid at all cost.

As to RC filtering, I intended to counter that by swapping the output caps for a pair of 47uF/6.3V Black Gate Nx. Am I correct in my calculation that for a TDA1543 powered with 8V power supply, maximum voltage seen by the output capacitor will be 5V (i.e. the voltage swing component: 8V - 1.8V - 1.2V = 5V)? I would hate to blow such rare caps by overvolting them.

If the output of the TDA1543 can`t get closer than 1.8 volts from the 8 volt supply the maximum voltage would be approximately 6.2 volts. This is if the output gets permanently digitally set to maximum. The other factor is that for any reasonable size of external resistor you select after the capacitor will cause the breakdown current to be limited by that resistor, probably in the uAmp region. In other words more leakage would appear as a voltage drop across your external resistor that in turn reduces the remaining voltage across your capacitor. If you have a voltage drop of any significance across your external resistor this will indicate you have some leakage issues.

The effect of externally loading the output with a lower value resistor is intimately connected with the value of the resistor connected between the output of the TDA1543 to ground. This is the resistor before the output capacitor. For example if both the external resistor and this internal resistor are 1kohm the output will drop to 50%. This is because the current swing of the DAC doesn`t change and the DAC sees the parallel combination of the two resistors as 500 ohms resulting in 50% swing. You could attach 10 ohm on the output, this resulting in 1% swing if the internal resistor is 1kohm. There is only a reduction in volume, the dynamic range actually increases if the DAC could produce more current.

What is interesting to note is that a current source with a 1k ohm load can be equally modeled as a voltage source with a series 1k ohm source resistance.
This means that when you parallel up a number of TDA1543 chips the equivalent source impedance of the DAC drops to the value of the resistor to ground needed to set up the quiescent operating point. This helps in lowering the source impedance of the DAC when directly driving cables and may account for some of the audible differences in using multiple DAC chips.

[uncle_leon]

I have just tested a number of resistors at my amp's input. It has resistor sockets so it was a relatively quick and easy job. I decided to run these tests before swapping the coupling caps.

I started with 3.48K, and haven't noticed any problems at all. Only the volume went up considerably - I use a passive, resistor-based "preamp" so I understand the decrease in DAC output was more than offset by a lower voltage divider ratio at the preamp.

I then tried subsequently lower values: 2.5K, 2K, 1K, and finally 300R. I currently use 4.7uF BG N coupling caps and the bass filtering only became obvious at 300R - admittedly, I only tested it with classical, might have been more apparent on heavier material.

I think the low-level detail became a little clearer with the lower resistors (1K and 300R). I am not sure about the dynamics as the changes in volume made it all hard to judge, but I think it was largely the same. All in all, the results are encouraging, and I will certainly be experimenting further.

If the output of the TDA1543 can`t get closer than 1.8 volts from the 8 volt supply the maximum voltage would be approximately 6.2 volts. This is if the output gets permanently digitally set to maximum.

Yes, but is it not 6.2V relative to ground?

Please correct me if I'm wrong, but the way I understand it is that when there is no signal present, the output stays at the "voltage midpoint", which for me calculates as 1.8V + (5V / 2) = 4.3V. When there is signal present, voltage can deviate by a maximum of 5V/2 =2.5V either way from this midpoint. So, relative to the output cap, the midpoint becomes a "0V", and the only voltage actually seen by that cap, is the +/- 2.5V signal swing.

[Hierfi]

Originally Posted by uncle_leon

I have just tested a number of resistors at my amp's input. It has resistor sockets so it was a relatively quick and easy job. I decided to run these tests before swapping the coupling caps.

I started with 3.48K, and haven't noticed any problems at all. Only the volume went up considerably - I use a passive, resistor-based "preamp" so I understand the decrease in DAC output was more than offset by a lower voltage divider ratio at the preamp.

I then tried subsequently lower values: 2.5K, 2K, 1K, and finally 300R. I currently use 4.7uF BG N coupling caps and the bass filtering only became obvious at 300R - admittedly, I only tested it with classical, might have been more apparent on heavier material.

I think the low-level detail became a little clearer with the lower resistors (1K and 300R). I am not sure about the dynamics as the changes in volume made it all hard to judge, but I think it was largely the same. All in all, the results are encouraging, and I will certainly be experimenting further.

Yes, but is it not 6.2V relative to ground?

Please correct me if I'm wrong, but the way I understand it is that when there is no signal present, the output stays at the "voltage midpoint", which for me calculates as 1.8V + (5V / 2) = 4.3V. When there is signal present, voltage can deviate by a maximum of 5V/2 =2.5V either way from this midpoint. So, relative to the output cap, the midpoint becomes a "0V", and the only voltage actually seen by that cap, is the +/- 2.5V signal swing.

6.2 volt is relative to ground. My calculations are for worst case as it is usually best to consider what could happen. If all is good under such circumstances any normal quiescent operating point doesn't matter. In any event I don't see a problem under the worst case condition possible.

Some further issues that you could run into is when using high value capacitors in conjunction with low values of resistance after the capacitor. If you turn the DAC off, when the power supplies falls on the DAC a high value current can feed back through the DAC's output into its power supply and damage the DAC itself. Protection against this can be found in some regulators that call for a reverse diode to bypass this current.

The worst case of risk to dynamic overload is in using the highest value of resistance. This generates the highest voltage. This results from the fixed current swing of the DAC feeding into a fixed resistance, or when V output equals (I x R) and I is a constant.

If you are reducing the value of the resistor at the amplifier end you are also reducing the source impedance as seen by the amplifier, or alternatively the output source impedance of your resistive divider preamplifier into the amp. By adding low value resistors this also upsets the natural logarithmic nature of the resistive divider as set by the manufacturer. This doesn't matter if you can still find settings to your liking. The clarity your are finding could be the result of lowering source impedance as seen by your amplifier.

The other consideration is that the resistive load seen by your DAC is also variant because of adding a low output resistor on your preamp. In other words if you set the preamplifier to its maximum output this transfers the 300 ohm resistor to the input of your resistive divider preamp, this then appearing as a load on the output of your DAC. Under such circumstances when using 300 ohms the low frequency extension becomes limited in relation to the coupling capacitor seeing this variant load. Ultimately your low frequency becomes limited as the gain setting on your preamp is increased. If this isn't at absolute maximum this shouldn't be a problem either.