Hi tessier,
Here are some basic properties for a TDA1541A I/V stage :
- Output compliance of +25 / -25mV at the DAC output -must- be met, these are absolute maximum values.
- Bandwidth of the TDA1541A output signal must be limited prior to feeding it into an I/V stage.
- All bit currents -must- be returned to +5V
- The I/V stage should be perfectly able to handle the presented bandwidth.
- Full DC-coupling is -required-
- I/V stage should be able to resolve signals down to 61 nano amperes / 30 microvolts.
Your circuit generates -0.004 * 35 = -140mV at the DAC output, this grossly exceeds output compliance. Since no +2mA bias current is used, signal does not swing around 0V, this increases the problem.
When using a suitable +2mA bias current source, max. I/V resistor value would be: 0.05 / 0.004 = 12.5 Ohms. If the +2mA bias current is not used, the I/V resistor value has to be reduced to 6.25 Ohms.
The bandwidth of the TDA1541A output signal is too large to directly drive a passive I/V resistor or circuit. So the output signal must be bandlimited prior to feeding it to an I/V resistor or active circuit. This can be done by connecting a suitable capacitor between TDA1541A output and GND or connecting a suitable capacitor at the output of a step-up transformer.
Step up transformers can limit bandwidth, provide sufficient signal amplitude while meeting output compliance, and also handle low level signals quite well.
Suppose we use a Sowther 9762 1:12.8 step-up transformer. When using a 1K I/V resistor at the transformer secondary, the reflected secondary load will be 1000 / 12.8^2 = 6.1 Ohm. This will meet output compliance (-0.004 * 6.1 = -24.4mV). The output signal amplitude will then be 312mVpp. If an output signal of say 6.65V is required, tube amplifier gain needs to be 18.
It is also possible to use both secondary windings of the 9762 to drive a differential tube amplifier stage. In this case we have 2 I/V resistors of 500 Ohms each.
This however doesn't solve the issue with bit return currents. The selected bit currents are routed through the reflected load of 6.1 Ohm to GND. The unselected bit currents are routed to +5V (inside the TDA1541A). This means that the load current on the +5V varies with the signal, this results in a ripple voltage that is extremely difficult to reduce to required low levels in the microvolt range. The ripple on the +5V (that also powers logic circuits) will introduce trigger uncertainty, introducing deterministic jitter (jitter related to music content).
The use of a transformer no longer provides full DC-coupling, full DC-coupling is required for highest transparency.
Here are some basic properties for a TDA1541A I/V stage :
- Output compliance of +25 / -25mV at the DAC output -must- be met, these are absolute maximum values.
- Bandwidth of the TDA1541A output signal must be limited prior to feeding it into an I/V stage.
- All bit currents -must- be returned to +5V
- The I/V stage should be perfectly able to handle the presented bandwidth.
- Full DC-coupling is -required-
- I/V stage should be able to resolve signals down to 61 nano amperes / 30 microvolts.
Your circuit generates -0.004 * 35 = -140mV at the DAC output, this grossly exceeds output compliance. Since no +2mA bias current is used, signal does not swing around 0V, this increases the problem.
When using a suitable +2mA bias current source, max. I/V resistor value would be: 0.05 / 0.004 = 12.5 Ohms. If the +2mA bias current is not used, the I/V resistor value has to be reduced to 6.25 Ohms.
The bandwidth of the TDA1541A output signal is too large to directly drive a passive I/V resistor or circuit. So the output signal must be bandlimited prior to feeding it to an I/V resistor or active circuit. This can be done by connecting a suitable capacitor between TDA1541A output and GND or connecting a suitable capacitor at the output of a step-up transformer.
Step up transformers can limit bandwidth, provide sufficient signal amplitude while meeting output compliance, and also handle low level signals quite well.
Suppose we use a Sowther 9762 1:12.8 step-up transformer. When using a 1K I/V resistor at the transformer secondary, the reflected secondary load will be 1000 / 12.8^2 = 6.1 Ohm. This will meet output compliance (-0.004 * 6.1 = -24.4mV). The output signal amplitude will then be 312mVpp. If an output signal of say 6.65V is required, tube amplifier gain needs to be 18.
It is also possible to use both secondary windings of the 9762 to drive a differential tube amplifier stage. In this case we have 2 I/V resistors of 500 Ohms each.
This however doesn't solve the issue with bit return currents. The selected bit currents are routed through the reflected load of 6.1 Ohm to GND. The unselected bit currents are routed to +5V (inside the TDA1541A). This means that the load current on the +5V varies with the signal, this results in a ripple voltage that is extremely difficult to reduce to required low levels in the microvolt range. The ripple on the +5V (that also powers logic circuits) will introduce trigger uncertainty, introducing deterministic jitter (jitter related to music content).
The use of a transformer no longer provides full DC-coupling, full DC-coupling is required for highest transparency.
So are you saying that the only sufficient I/V stage for a TDA1541 is power by the same +5V for the chip? That would severely limit any potential I/V stage design. This just doesn't sound reasonable, basically our only option would be an IC opamp or a terribly high distorting jfet buffer.
Hi
I've seen in the thread that some guy's suggested to use the Aikido tube preamp as a IV amp.
Is it a good ideas and could you tell me where I can find a IV amp circuit using the Aikido ?
I have few 12AU7 to made a IV amp.
Thanx a lot.
Paul
Last edited:
ECC82 is good in general for amplifying the low signal after the R-IV.
iF You dont mind the higher output resistance of 6K cca. caused with 6.5K-7K Internal resistance of the tube? In a simple basic topology.
And phases will be good too because R-iv shifts of 180deg and Tube stage shifts too of 180 deg...
But this circuit from post 4701 is complicated, no additional phase shift,
tooo many C on signal path, AND very low anode voltages 40V cca per tube section...
Tis is tube deserves higher voltage, not a transistor.
there is no need for that, but just for plain circuit.
cheers
iF You dont mind the higher output resistance of 6K cca. caused with 6.5K-7K Internal resistance of the tube? In a simple basic topology.
And phases will be good too because R-iv shifts of 180deg and Tube stage shifts too of 180 deg...
But this circuit from post 4701 is complicated, no additional phase shift,
tooo many C on signal path, AND very low anode voltages 40V cca per tube section...
Tis is tube deserves higher voltage, not a transistor.
there is no need for that, but just for plain circuit.
cheers
I think with an I/V resistor this low the only chance would be a Jfet phopreamp type gain stage.
With a tube the SNR would kill resolution.
One of the quietest tubes is a triode strapped D3A and it has nice gain but even it would have trouble amplifiying a .002mAx10 ohms*0.0015849 LSB = 0.03 micro Volt signal.
Probably end up with a 12 bit resolution DAC at best with a 10 ohm I/V resistor.
Oleg did some measurements and found that a 33 ohm resistor with the 2mA offset handled as Pedra does was the best compromise for a tube gain.
Hi tessier,
Here you can find information about Aikiodo amplifiers and kits:
GlassWare Line Stage & Headphone Amplifiers Kits and PCBs
Aikido LV
Here you can find information about Aikiodo amplifiers and kits:
GlassWare Line Stage & Headphone Amplifiers Kits and PCBs
Aikido LV
I tried it. The chip delivers constant current into a short circuit. However I noticed that the chip is rather hot (even during normal load, not just with shorted output). So I glued a copper heatsink on top in order to reduce the surface temperature, reduce vibration (might be woodoo, but costs nothing), and reduce EMC sensitivity. I grounded the heatsink with a piece of wire to DGND.
Hi Zoran,
TDA1541A is based on current mode logic / transistors, CML draws a constant current, this is why the TDA1541A gets hot. The advantage of CML is that supply current remains the same during switching or during steady state. This results in low on-chip switching noise.
Modern DAC chips are based on CMOS logic, CMOS logic only draws current during switching as stray capacitance needs to be charged / discharged. These chips run much cooler, but the large differences in supply current between steady state and swithing leads to high on-chip switching noise.
TDA1541A Typical power dissipation: 700mW
TDA1541A Maximum power dissipation 1W
Power dissipation doesn't change with load impedance.
The same as without load, in free air @ 19 dedrees Celcius room temperature, chip surface temperature (center of the chip) is around 40 degrees Celcius (typical). Chip temperature can rise significantly when the chip is placed in a closed housing without sufficient (convection) cooling and elevated ambient temperature inside that housing.
TDA1541A is based on current mode logic / transistors, CML draws a constant current, this is why the TDA1541A gets hot. The advantage of CML is that supply current remains the same during switching or during steady state. This results in low on-chip switching noise.
Modern DAC chips are based on CMOS logic, CMOS logic only draws current during switching as stray capacitance needs to be charged / discharged. These chips run much cooler, but the large differences in supply current between steady state and swithing leads to high on-chip switching noise.
TDA1541A Typical power dissipation: 700mW
TDA1541A Maximum power dissipation 1W
Power dissipation doesn't change with load impedance.
The same as without load, in free air @ 19 dedrees Celcius room temperature, chip surface temperature (center of the chip) is around 40 degrees Celcius (typical). Chip temperature can rise significantly when the chip is placed in a closed housing without sufficient (convection) cooling and elevated ambient temperature inside that housing.
Can this be applied to the TDA1543?
what would be the bias currents? Possible to avoid the -15V rail?
Last edited: