At a few MHz it might make me happier, at a few GHz, it wouldn't cut it.Interesting approach. I think that's a bit overkill, but if it makes you happy...
Tom
At audio frequency? everthing to do with ground needs a lot of thought for me!
I'm sure that's fun for some people, but some others want to do more than assembly work.
Around here, you can get lots of Cambridge Audio A1's to fix up much cheaper than buying a kit. Could be a cheap way to learn a lot for some people.
I'm not knocking the satisfaction of putting something together, or the performance of anyone's work, just questioning what people want from the process.
I found a kit was the best way to start in 1980 when I knew very little about electronics. Learning to recognise transistors, resistors etc.
Learning to read a power supply wiring diagram.
Then wondering what on earth I did wrong when it doesn't work.
Then reading up on amplifiers to get a better understanding of how they work.
37 years on I am designing my own amps but very much still learning when things don't work quite right.
I think that's a terrific idea!! You can get all the bits you need for a complete amplifier (chassis, switching, power supply, etc.) and then just customize it. Looking at the A1 schematic, though, I'm wondering how they get away with NO coupling capacitors whilst using NE5532s.
Presumably, they are convinced there is no DC on the input?
But in the schematic I have, there are caps in the lines from the volume pot.
AIUI, there are quite a few versions of this amp. I've had one for around 20 years, I think it sounds OK, subjectively maybe it's better with my old 16ohm speakers? It's been demoted and replaced by an Arcam, simply because that drives two pairs of speakers.
I picked up another broken A1 on ebay for about twelve quid inc P&P.
Are there any versions of the Cambridge Audio A1 that have dual (110/220v) power transformer primaries?
+1 to daniel's vote for Tom's work. There's more than enough reading on his site wherein he explains the extra pains he goes through to "tame the 3886", to give a newbie a good foundation, and I can attest from personal experience to the results. Opting for the unpopulated boards and BOM gives the builder a chance to learn more than just the chassis bashing and plugging in of modules as you might find with the DIY Class D (Hypex), or as in my own case Sure prebuilt boards and MeanWell SMPS .
DC servo opamp should be FET input.
A 5534/5532 is not well suited.
You can use the ancient LF411 (single) or LF412 (dual) for DC servo duty.
There are very many modern fet input opamps that have better specifications than the 411/412 but they all cost more.
If the output from the DC servo has no audio content, then it can't affect the audio quality. That probably requires outputs at >2Hz to be virtually unmeasurable.
The important parameters are supply voltage, input offset voltage and drift of input offset voltage.
A 5534/5532 is not well suited.
You can use the ancient LF411 (single) or LF412 (dual) for DC servo duty.
There are very many modern fet input opamps that have better specifications than the 411/412 but they all cost more.
If the output from the DC servo has no audio content, then it can't affect the audio quality. That probably requires outputs at >2Hz to be virtually unmeasurable.
The important parameters are supply voltage, input offset voltage and drift of input offset voltage.
Thanks for the answer.
I have a couple of OPA1642's at home, those are fet input opamps. I also have a few OPA2107 and OPA2134.
For supply voltage I was planning on using 2*24vac toroidal transformer, MUR2040 and 4*10000uF. In other words, not use the onboard rectifier.
I could use a pair of LT1083's if I would benefit at all from having regulated supplies.
I assume that the DC-servo (opamp) has it's own regulated supplies taken from the LM3886 supplies.
My point about the 24VAC transformer was that I will power those rails with it and an external capacitor bank (leaving out the onboard rectifier bridge) and use MUR2040's before the external cap bank.
The LF411 is not bad for its age and certainly more suitable for DC servo duty than the NE5534/5532. The 500 uV offset voltage for the A grade (2 mV for the non-A) was decent back then. The killer is that the input bias current gets significant as the device heats up. 4 nA at 70 ºC and 50 nA at 125 ºC. That's the leakage current of the input JFET you see there. The issue is that this input current will cause an additional offset voltage.
Note that the LF411ACN from National Semiconductor (now TI) has been obsoleted by TI a few months ago. The corresponding TI part is not as well specified.
A better choice would be a MOSFET input opamp, such as the OPA277. It's twice the cost of the LF411 but also much better spec'ed. 50 uV input offset, 2 nA input bias current, 2 nA input offset current.
All numbers above are the worst case (max) specs.
True.
I'd argue 20 Hz, but whatever. The key here is the "immeasurable" part. There are two ways to achieve that.
- Push the cutoff frequency far enough down that content at 20 Hz becomes immeasurable. Unfortunately this also maximizes the settling time of the DC servo.
- Use a steeper filter slope on the DC servo. This increases component count but reduces the settling time of the DC servo.
You can determine the impact of the DC servo from simulation or from measurement of the THD at 20 Hz. The stop band attenuation of the DC servo needs to be high enough that there is no degradation of the THD at 20 Hz when the DC servo is engaged.
That was certainly true 20-30 years ago. Nowadays, it's trivial to find an opamp with < 1 mV of offset and reasonable low drift that can handle ±18 V or more. The key challenges with modern parts are low input current (both bias and offset) and low noise.
The DC servo opamp does need to be reasonably low noise or you'll see the impact of its noise within the audio band of the amplifier. Thankfully, low current noise comes with the MOSFET inputs, so you "just" need to find a MOSFET input opamp with low voltage noise. I'd aim for 10 nV/rtHz or below with as low 1/f corner as you can get. I'd also simulate the circuit to ensure that the noise of the DC servo does not impact the performance of the overall amp.
DC servo design is a topic of its own, in particular if you want to push the envelope of performance.
Tom
Last edited:
I do have one or two of the OPA2107, but I meant OPA2111KP. Those I have a few of.
IIRC, I've read somewhere on here that you want a low noise opamp as DC-servo. That was in a different application though (headphone amplifier IIRC).
Hmmm....now an OPA1642 has LESS input bias & offset current (20pA max), but more input offset voltage (3.5mV max). So which is more important?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.