Hi,
I am currently working on a balanced floating output. My current prototype is based on two compensated 5534 as shown in the attached png file. It will drive a 10kohms electronically balanced input stage at line level (+4dBu). The objective is to get a buffer stage as neutral as possible.
An alternative could be to replace the two 5534 by a single 5532. The 5534 is known to be better sounding than the 5532. However a single 5534 would reduce the area of the PCB (and its number of components and its final cost).
Did any of you compared the sound of a similar output stage based on two 5534 to the "equivalent" stage based on a single 5532 ?
I am currently working on a balanced floating output. My current prototype is based on two compensated 5534 as shown in the attached png file. It will drive a 10kohms electronically balanced input stage at line level (+4dBu). The objective is to get a buffer stage as neutral as possible.
An alternative could be to replace the two 5534 by a single 5532. The 5534 is known to be better sounding than the 5532. However a single 5534 would reduce the area of the PCB (and its number of components and its final cost).
Did any of you compared the sound of a similar output stage based on two 5534 to the "equivalent" stage based on a single 5532 ?
Attachments
The servos are to compensate a DC offset at the output, ideally to avoid caps in series. However I also plan to test the impact of caps at the output (with Nichicon Muse 22uF, optionally with 10nF polystyrene cap in parallel) but I do not expect this option to be neutral. I assume that the driven gear has electronically balanced inputs but it may be used to drive xmfr. Why would series C be safer than compensated DC ?
Regarding the sonic difference between a dual 5534 and a single 5532 driving the balanced output, I would be interested in your conclusions if any of you already made the test with a similar topology.
Regarding the sonic difference between a dual 5534 and a single 5532 driving the balanced output, I would be interested in your conclusions if any of you already made the test with a similar topology.
tle2072 isn't any lower Vos than the 5534 and the DC gain is only 2x - you need another op amp for the the servo to meaningfully reduce Vos at the output - and another circuit to preserve cmrr balance while injecting the DC offset adj
with the 5534 singles for the diff amps you could add Vos trimpots and skip the servos - connect the "ref" legs to gnd per the usual diff amp circuit
if the source can drive the unbuffered diff amp resistors why do you need a "buffer" with a 10K load?- just series caps removes DC, bigger value, bipolar construction, high V rated Al electros give pretty low measured distortion in Bateman's articles
I think its too hard, slow, expensive to do the properly controlled blinded listening tests required to turn subjective circuit "sound" perceptions into actionable engineering recommendations
My answer is to design for the best possible technical performance with the latest op amps selected for their technical specs in the application circuit and to employ more complex topologies where there is engineering justification to believe they offer superior technical performance
there seems to me to be little point in spending hrs trying to optimize 50 cent op amps when I can buy best in class for the application for $5 – I certainly value my hobby time at a higher rate than that
with the 5534 singles for the diff amps you could add Vos trimpots and skip the servos - connect the "ref" legs to gnd per the usual diff amp circuit
if the source can drive the unbuffered diff amp resistors why do you need a "buffer" with a 10K load?- just series caps removes DC, bigger value, bipolar construction, high V rated Al electros give pretty low measured distortion in Bateman's articles
I think its too hard, slow, expensive to do the properly controlled blinded listening tests required to turn subjective circuit "sound" perceptions into actionable engineering recommendations
My answer is to design for the best possible technical performance with the latest op amps selected for their technical specs in the application circuit and to employ more complex topologies where there is engineering justification to believe they offer superior technical performance
there seems to me to be little point in spending hrs trying to optimize 50 cent op amps when I can buy best in class for the application for $5 – I certainly value my hobby time at a higher rate than that
The attached png is only to illustrate an balanced floating output with servo. It is not the complete design, which functionality is not a only a balanced buffer. I considered the other alternatives to the servos, skipping them is not an option. Regarding the choice of the TLE2072, I have seen (heard) very good solutions with technically worse op-amps such as LF353. 
My question is to know whether someone tried to compare the sound of a 5532 vs dual 5534 for a balanced floating output stage. Some claim that the 5534 are more detailed than the 5532, the other point could have been the channel separation of the 5532 which I would expect not to be an issue for this use. Up to what point it worth to use two 5534 rather than a 5532, say from a sound engineer perspective looking for neutrality.
My question is to know whether someone tried to compare the sound of a 5532 vs dual 5534 for a balanced floating output stage. Some claim that the 5534 are more detailed than the 5532, the other point could have been the channel separation of the 5532 which I would expect not to be an issue for this use. Up to what point it worth to use two 5534 rather than a 5532, say from a sound engineer perspective looking for neutrality.
the servo op amp should have much less Vos than you are trying to correct - often seen with discrete amps with large DC gain where the un-adjusted Vos at the output may be 100s mV - or even Volts for phono preamps
using 5 mV Vos spec servo op amps to "fix" 5mV Vos diff amp with DC gain of 2 isn't a good option in my estimate
the only reason I see to have 2 servos is if you also require diff mode high pass
if what you really want is a range of opinions on 5534 see: http://www.diyaudio.com/forums/solid-state/62806-5534-audio-amp.html#post706550
using 5 mV Vos spec servo op amps to "fix" 5mV Vos diff amp with DC gain of 2 isn't a good option in my estimate
the only reason I see to have 2 servos is if you also require diff mode high pass
if what you really want is a range of opinions on 5534 see: http://www.diyaudio.com/forums/solid-state/62806-5534-audio-amp.html#post706550
Last edited:
I conducted some tests of my prototype in a recording studio yesterday. The prototype driving a balanced input of a Metric Halo, routed to K&H O300D monitor speakers.
Surprisingly, the TI 5532P driving the hot&cold outputs produced a slightly more natural sound than two compensated TI 5534AP.
On the servo side, I compared a 1st order to a 2nd order (MFB) low-pass filter at 1Hz (with 0.33u MKT caps), the 1st order LPF produced a slightly more neutral in the highs for a very slight loss of low end bass than the 2nd order. [by slight, I mean close to the feeling one have when comparing a well encoded high-rate mp3 to its 24-bit master]
Surprisingly, the TI 5532P driving the hot&cold outputs produced a slightly more natural sound than two compensated TI 5534AP.
On the servo side, I compared a 1st order to a 2nd order (MFB) low-pass filter at 1Hz (with 0.33u MKT caps), the 1st order LPF produced a slightly more neutral in the highs for a very slight loss of low end bass than the 2nd order. [by slight, I mean close to the feeling one have when comparing a well encoded high-rate mp3 to its 24-bit master]
I read the same post and bought a few Brown Dogs to check the difference between 5532 / 5534 myself. Conclusion: Very little difference. Probably it's all depends on the circuit.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.