Hi, I am designing a high current driver after a precision DAC (AD5781). I found the circuit from AN18: Power Gain Stages for Monolithic Amplifiers. In Figure 02, It uses a current boost design of NPN + PNP after a high current buffer LT1010, which is capable of driving 150 mA output. The overall circuit can deliver 3A RMS current at +, - 15V load.
We have an output buffer AD8675 after the 18 bit DAC AD5781, and we want to use the above circuit after the DAC for driving a high current load. We have the following questions,
1, is this circuit requires a special op-amp buffer such as LT1010? For example, in this circuit, the 33 Ohm to the VSS, VDD and 100 Ohm from the output of LT1010 to the ground provide biasing of Q1, Q2. Does this mean that Q1, Q2 are always on and there are minimal crossover distortions (high precision)?
2, Can we replace LT1010 with a more precision high current amplifier such as ADA4870 or LT1970A? If we can, what should we take care of in order to make this circuit stable?
3, Can we replace LT1056 and LT1010 altogether with a modern op-amp? for example, ADA4627-1, AD8675 or LT6018? We like the specification of LT6018, and would like to know whether LT6018 (AD8675) alone or LT6018 (AD8675) + ADA4870 can replace LT1056 + LT1010?
4, How to analysis the overall performance (noise, precision, temperature drift)? Is this power booster circuit preserves 18-bit precision?
1. The function of the LT1010 in that circuit is to take the current in the 100 Ohm load and put the current into either of the 33 Ohm resistors in its supply. At low currents (below about 18mA), Q1 and Q2 are off. As the current rises, the voltage across either of the 33 Ohm resistors rises above 0.6V, turning on Q1 or Q2. [Simulations are not very helpful with circuits like this, because the supply currents are not necessarily correctly modeled.]
2. There's not much point in replacing the LT1010 with a precision opamp, because the precision is provided by the LT1056. Generally speaking, it is VERY difficult to get precision and power in the same chip. The on-chip thermals ruin the precision every time. But since this circuit was made, we have released higher current drivers such as LT1210 and ADA4870. So you could replace the LT1010 and the BJTs with a single high power opamp, or multiple high power opamps more or less in parallel. And the trick is as you suggest, getting the compensation right. In the circuit included, that's the function of the 15pF and the 68pF.
3. Yes, other precision+power circuits are certainly possible. The LT6018 shows one in the Wien Bridge app circuit, for example, using an LT1206 in the loop for the higher load driving capability.
4. The input stage dictates the noise and precision. It needs a high gain to overcome any errors in the boost circuit. Because the input stage is inverting, the LT1056 will not have any common mode issues. But its tempco of 4uV/C could impact 18 bit performance. Also, at that level of performance you would want to consider the voltage coefficient of the main gain resistors (10k:10k in this case).
I am still confused about the 33 Ohm and 100 Ohm settings.
If Q1 is pre-biased, then current flow through 33 Ohm must be greater than 0.6V, therefore, the voltage drop from +15V to VCC pin of LT1010 should be more than 0.6V. I found the supply current is maximum 9mA/10mA, so the voltage drop over 33 Ohm should be 0.33V, right? In your reply, what is 18 mA comes about? And what are the function of 100 Ohm output to ground and another 100 Ohm ties to the C of Q1, Q2?
I want to use this circuit for both power and minimal crossover distortion, I do want to simulate this circuit in LTSpice but if this circuit cannot be easily simulated, how could I evaluate the levels of crossover distortion?
Can I replace LT1010 with ADA4870 where both MJE2955 and MJE3055 are kept and 33 Ohm and 100 Ohm connections are the same as in this circuit. Will it work for a current boost circuit?
Q1 is off at quiescent. It only turns on as the LT1010 tries to pull up on the 100 Ohm load, which then pulls down on the 33 Ohm.
The ADA4870 may work, but it would handle more of the load so you'd have to scale back the 33 Ohms. Also, you wouldn't wnat to waste 1A in that dummy 100 Ohm load, so the topology is not best suited to ADA4870.
1. 18mA * 33Ohm = 594mV, so Q1 is on if 18 mA is flowing through 33Ohm (+15V to VCC). You mention that the output of LT1010 should be non-zero. Does this mean that the minimum output voltage of LT1010 should be at least 1.8V for turning on Q1? There is still some crossover distortions, right?
2. Can I replace the combination of (LT1056 + LT1010) with a single op-amp, such as LT6018? The Iout (typical value) of LT6018 is 90mA, Can I use LT6018 in this way?