I'm very interested in getting the forum expert's technical critique of this IMO slightly unusual high end production amplifier from the late 90's. Design highlights are as follows. High bias (class A likely close to 10 watts), high power bandwidth to 150khz, high input impedance at 150K, doubles power output from 8 to 4 ohms, low 6db feedback, and is absolutely stable with no output filtering network. The amp also has 130,000uf of overall power supply filtering.
Q15 & 16 in the servo circuit is the only issue I find, in that they require tight matching otherwise audible 60hz hum is introduced.
Though it may not measure impressively well with regards to THD and IM distortion, this is one of the most dynamic and musical 100 watt A/B amplifiers I have ever heard. Presenting a big easy sound that is never harsh, while still being wonderfully transparent and detailed. All this in a well built compact (37 pound) all aluminum chassis.
Thoughts?
Q15 & 16 in the servo circuit is the only issue I find, in that they require tight matching otherwise audible 60hz hum is introduced.
Though it may not measure impressively well with regards to THD and IM distortion, this is one of the most dynamic and musical 100 watt A/B amplifiers I have ever heard. Presenting a big easy sound that is never harsh, while still being wonderfully transparent and detailed. All this in a well built compact (37 pound) all aluminum chassis.
Thoughts?
Last edited:
I omitted the manufacturer/model to remove any preconception from the conversation. Apologizes if this was needed.
ARC has had good and bad products, like any other company.
The first D100 model from 1977 was certainly not very good.
The first D100 model from 1977 was certainly not very good.
The amplifier is designed for good open-loop linearity with little use of negative feedback.
This approach has two downsides:
The first may not be noticeable in normal listening. The second likely accounts for the amplifier sounding "different".
Ed
This approach has two downsides:
- Distortion in the class AB output stage will be high at power levels that cause the output transistors to cut-off.
- The output impedance may be high enough to affect the speaker crossover.
The first may not be noticeable in normal listening. The second likely accounts for the amplifier sounding "different".
Ed
Agreed rayma. ARC's claim to fame have always been their vacuum tube offerings. This 100.2 is their last solid state A/B design before going class D, It represents a solid progression from previous attempts.
Yes. The distortion is low as long as the output transistors remain in class A.
Ed
Ed
That might explain the over-abundance of FETs! And the high impedances (SNR will show the effect of this). Seems very over complex front end - be interesting to hear the logic behind such a design (some of it is the support for balanced input I think). 56pF isn't great protection against RF ingress though...
Mark Tillotson, The amp is very quiet aside from a slight audible 60/120hz hum due to Q15 & Q16 matching having drifted over the last 20 years.
Summing the current at the folded cascode junction rather than a current mirror is interesting.
It looks slightly more conventional when you ignore Q5 and Q8, then they simply take the output current from one side of each input differential pair. Q5 and Q8 are needed because of the differential input, so the differential-to-single-ended conversion and input differential pair are intertwined.
Without overall feedback, the DC servo would have a bandwidth of roughly 1/6 rad/s, 0.026 Hz, so it would take about 30 seconds to settle within 1 % of the initial offset. The overall feedback probably makes it even slower, and it gets slower again when you push the mute button.
Without overall feedback, the DC servo would have a bandwidth of roughly 1/6 rad/s, 0.026 Hz, so it would take about 30 seconds to settle within 1 % of the initial offset. The overall feedback probably makes it even slower, and it gets slower again when you push the mute button.
Just from the feedback network it must be 50µVrms or more - The feedback netwoek 162k:3k3 could easily be changed to a more conventional 47k:1k for an improvement, unless the FETs are dominating.
Despite the apparent complexity of the amplifier topology, it is a single-stage complementary amplifier, i.e. just one voltage amplification stage.
This is neither good nor bad, it is rather a compromise of characteristics. In the modern understanding, it is almost impossible to make a high-quality amplifier using a single amplification stage. But an amplifier for listening at home is quite suitable and will sound very decent.
For a push-pull output stage on bipolar transistors, a high initial current is not a synonym for quality, because uncompressed musical material has many peaks that the amplifier must play out efficiently, the transition of the output stage from A to class AB at high power is sometimes noticeable by ear and the quality is not adds.
It is not the feedback depth that is important, but the linearity of the stage gain before introducing feedback or loading with a low-resistance resistor, 6 dB is the excess gain that goes into the feedback, the linearity of the amplifier is ensured by the load resistor R45, but at the same time it worsens the control of the output stage, because The output stage consists of bipolar transistors.
This is a mutually exclusive comment. IM distortion is bad dynamics.
good sound here is the use of J-fet transistors at the input; they have a larger range of input voltages, so the resulting sound is cleaner.
The one rare thing I see is the DC servo adjusting the FET CCS's - neat ... I do this on both my symmetric designs. Otherwise , too many devices !
OS
If the DC servo really is as shown, I would expect U1 output to be stuck on a supply rail. R49 should be connected to ground rather tied at R50, C30.
BSST, would this alleviate the tight matching requirement between Q15 & Q16 to keep hum in check, while still allowing the servo to zero DC offset?