Good active crossovers, aside from the filters used, are usually designed to be unity gain devices as well as being very low noise.
In a standard passive loudspeaker white noise (hiss) is far less of an issue than with active. The low-pass crossovers, on midrange and bass drivers, automatically filter out the high frequency portion reducing its audibility. Not to mention the drivers themselves rolling off. Tweeters are normally the only problem but these tend to be attenuated by the crossover too reducing the severity of the problem again.
As active removes the passive crossover low noise, in the signal chain, becomes even more critical.
Just as with a passive crossover the signal chain within the active crossover can filter out excess high frequency noise. For example if your DAC was very noisy the active low-pass filter, on the bass section, would filter out some of this noise and make it less of an issue.
As you're reporting excessive amounts of hiss from all the drivers in the system it implies that the active crossover is an, intrinsically, high noise device. Looking at the schematics for it this isn't a surprise 10k/5k seems to be the go-to resistor size for feedback and input resistors.
As others have mentioned here the voltage swing that the active crossover can operate over is large. This allows it to work with very large signals. As the noise is high the only way for it to have a decent signal to noise ratio is to increase the signal level.
Typical 0dBFS single ended outputs from DACs are 2VRMS and balanced this doubles to 4VRMS. Even though the 10B offers balanced inputs and outputs it is not a fully balanced design. Internally it sums the balanced input to single ended, uses single ended filters, then converts this back into a balanced signal for the outputs.
Here is a review of the Fiio.
Review and Measurements of FiiO Q1 Mark II Portable DAC and Headphone Amp | Audio Science Review (ASR) Forum
We can see that at full scale, single ended, it is only outputting 1.5VRMS. This isn't as high as the typical standard but it isn't pathetically low either.
If you look at the manual for the Bryston amplifiers...
Page Not Found - Bryston[3B4B].pdf
Page Not Found - Bryston[7B].pdf
It shows that to reach maximum output power for the 3B, 4B and 7B requires significant voltage drive. This isn't a surprise as the amplifiers are very powerful and have a wide output voltage swing.
If you had the Fiio driving the 7B directly you'd need to use the Fiio in balanced mode (this will double its output swing. So instead of 1.5VRMS you'll be getting 3VRMS) with the 7B in high gain. Otherwise you'd never have enough voltage from the Fiio to get the amplifiers to their limits. Thankfully no one needs 700 watts for domestic listening, and no one ever wants their amplifiers to clip under normal operation either, so using them in low gain mode is a good way to minimise system noise and to help prevent over driving the amplifiers.
My advice here is to use the balanced outputs of the Fiio, to maximise it's output drive level and to then use the power amplifiers in low gain mode.
Now the question of the jumper positions inside the active crossovers and the stepped attenuators on the front that set the gain.
We don't want to be applying any gain within the crossover itself unless it's done in a very low noise gain stage. You also don't want to reduce the signal level passing through the active crossover, in any shape or form, either. The issue here isn't that the active crossover has a low signal to noise ratio, it's that it has a high level of intrinsic noise on its output. To achieve this high signal to noise ratio it needs a large signal passing through it. So we don't want any signal level reductions to occur inside.
As it stands it looks like you've got the thing configured pretty much for that already, which is both jumpers installed. I'm assuming the stepped attenuators are bypassed? Or are they functional?
Balanced the Fiio outputs ~3VRMS. Single ended the maximum output swing of the 10B is 13VRMS. The first thing the 10B does is sum a balanced input down to single ended, so the maximum signal level the 10B can pass through, without clipping, is 13VRMS.
The Fiio appears to have a signal to noise ratio of about 110dB and this is most likely referenced to 0dBFS but that doesn't mean this is where its noise floor ends. Most delta sigma DACs have a noise floor that falls as the signal level reduces.
You can see here
https://www.akm.com/content/dam/doc...dac/ak4452vn/ak4452vn-en-evaluationmanual.pdf
That under low signal levels the noise floor drops considerably. It's not that the signal to noise ratio of the DAC has increased but just that the noise has fallen with the signal level.
If you were to add in a gain stage after the DAC to boost the signal up to 13VRMS you'd still want it to be low enough noise so that the DAC is dominant under these low signal levels.
Either way to maximise the system signal to noise ratio you are going to want a low noise preamp after the Fiio applying 12.5dB of gain. Think a non inverting gain stage using OPA1612s with 600 ohm feedback resistors, or similar.
You would then want to have attenuators on the output of the 10B to reduce the signal level back down again to match the maximum volume level you want to listen at. Preferably about 20dB.
Analogue active crossovers are a pain though. To keep them very low noise you need to use low noise opamps in combination with low value resistors. The low value resistors necessitate large value capacitors making things take up lots more space and more costly. Not to mention the increased loading this places on the opamps too.
In an ideal situation, for an analogue active, you want to have the DAC always operating at 0dBFS without any digital attenuation. This feeds into the analogue active crossover with the crossover optimised around maximising its signal to noise ratio with respect to typical line level signals. 2VRMS for SE and 4VRMS for balanced. You then use a low noise multichannel preamp after the active crossover to set the volume. This means that for anything other than very loud listening levels hiss should be completely inaudible.
The fact that the 10B is a high noise part, feeding directly into the power amps, is the problem. Really what you need is a low noise multichannel preamp after it and a low noise gain stage before it.
The one thing that concerns me is when you said these things.
The inherent noise of semiconductors does not suddenly change at times unless there's something in error.
If you wanted to build your own active crossover that's always an option. You've already got the case for the 10B, all the input and output connectors, power supply etc, all you'd need to do is replace the active crossover boards and potentially regulate the power rails down to +-15/18V. It would be worth measuring the outputs of the 10B to see what the transfer function of the internal filters are. They might be very simple/generic and easy to replicate using something else and much lower noise.
In a standard passive loudspeaker white noise (hiss) is far less of an issue than with active. The low-pass crossovers, on midrange and bass drivers, automatically filter out the high frequency portion reducing its audibility. Not to mention the drivers themselves rolling off. Tweeters are normally the only problem but these tend to be attenuated by the crossover too reducing the severity of the problem again.
As active removes the passive crossover low noise, in the signal chain, becomes even more critical.
Just as with a passive crossover the signal chain within the active crossover can filter out excess high frequency noise. For example if your DAC was very noisy the active low-pass filter, on the bass section, would filter out some of this noise and make it less of an issue.
As you're reporting excessive amounts of hiss from all the drivers in the system it implies that the active crossover is an, intrinsically, high noise device. Looking at the schematics for it this isn't a surprise 10k/5k seems to be the go-to resistor size for feedback and input resistors.
As others have mentioned here the voltage swing that the active crossover can operate over is large. This allows it to work with very large signals. As the noise is high the only way for it to have a decent signal to noise ratio is to increase the signal level.
Typical 0dBFS single ended outputs from DACs are 2VRMS and balanced this doubles to 4VRMS. Even though the 10B offers balanced inputs and outputs it is not a fully balanced design. Internally it sums the balanced input to single ended, uses single ended filters, then converts this back into a balanced signal for the outputs.
Here is a review of the Fiio.
Review and Measurements of FiiO Q1 Mark II Portable DAC and Headphone Amp | Audio Science Review (ASR) Forum
We can see that at full scale, single ended, it is only outputting 1.5VRMS. This isn't as high as the typical standard but it isn't pathetically low either.
If you look at the manual for the Bryston amplifiers...
Page Not Found - Bryston[3B4B].pdf
Page Not Found - Bryston[7B].pdf
It shows that to reach maximum output power for the 3B, 4B and 7B requires significant voltage drive. This isn't a surprise as the amplifiers are very powerful and have a wide output voltage swing.
If you had the Fiio driving the 7B directly you'd need to use the Fiio in balanced mode (this will double its output swing. So instead of 1.5VRMS you'll be getting 3VRMS) with the 7B in high gain. Otherwise you'd never have enough voltage from the Fiio to get the amplifiers to their limits. Thankfully no one needs 700 watts for domestic listening, and no one ever wants their amplifiers to clip under normal operation either, so using them in low gain mode is a good way to minimise system noise and to help prevent over driving the amplifiers.
My advice here is to use the balanced outputs of the Fiio, to maximise it's output drive level and to then use the power amplifiers in low gain mode.
Now the question of the jumper positions inside the active crossovers and the stepped attenuators on the front that set the gain.
We don't want to be applying any gain within the crossover itself unless it's done in a very low noise gain stage. You also don't want to reduce the signal level passing through the active crossover, in any shape or form, either. The issue here isn't that the active crossover has a low signal to noise ratio, it's that it has a high level of intrinsic noise on its output. To achieve this high signal to noise ratio it needs a large signal passing through it. So we don't want any signal level reductions to occur inside.
As it stands it looks like you've got the thing configured pretty much for that already, which is both jumpers installed. I'm assuming the stepped attenuators are bypassed? Or are they functional?
Balanced the Fiio outputs ~3VRMS. Single ended the maximum output swing of the 10B is 13VRMS. The first thing the 10B does is sum a balanced input down to single ended, so the maximum signal level the 10B can pass through, without clipping, is 13VRMS.
The Fiio appears to have a signal to noise ratio of about 110dB and this is most likely referenced to 0dBFS but that doesn't mean this is where its noise floor ends. Most delta sigma DACs have a noise floor that falls as the signal level reduces.
You can see here
https://www.akm.com/content/dam/doc...dac/ak4452vn/ak4452vn-en-evaluationmanual.pdf
That under low signal levels the noise floor drops considerably. It's not that the signal to noise ratio of the DAC has increased but just that the noise has fallen with the signal level.
If you were to add in a gain stage after the DAC to boost the signal up to 13VRMS you'd still want it to be low enough noise so that the DAC is dominant under these low signal levels.
Either way to maximise the system signal to noise ratio you are going to want a low noise preamp after the Fiio applying 12.5dB of gain. Think a non inverting gain stage using OPA1612s with 600 ohm feedback resistors, or similar.
You would then want to have attenuators on the output of the 10B to reduce the signal level back down again to match the maximum volume level you want to listen at. Preferably about 20dB.
Analogue active crossovers are a pain though. To keep them very low noise you need to use low noise opamps in combination with low value resistors. The low value resistors necessitate large value capacitors making things take up lots more space and more costly. Not to mention the increased loading this places on the opamps too.
In an ideal situation, for an analogue active, you want to have the DAC always operating at 0dBFS without any digital attenuation. This feeds into the analogue active crossover with the crossover optimised around maximising its signal to noise ratio with respect to typical line level signals. 2VRMS for SE and 4VRMS for balanced. You then use a low noise multichannel preamp after the active crossover to set the volume. This means that for anything other than very loud listening levels hiss should be completely inaudible.
The fact that the 10B is a high noise part, feeding directly into the power amps, is the problem. Really what you need is a low noise multichannel preamp after it and a low noise gain stage before it.
The one thing that concerns me is when you said these things.
The inherent noise of semiconductors does not suddenly change at times unless there's something in error.
If you wanted to build your own active crossover that's always an option. You've already got the case for the 10B, all the input and output connectors, power supply etc, all you'd need to do is replace the active crossover boards and potentially regulate the power rails down to +-15/18V. It would be worth measuring the outputs of the 10B to see what the transfer function of the internal filters are. They might be very simple/generic and easy to replicate using something else and much lower noise.
Hi there, thanks a lot for your reply! You put a lot of thought into that.
BTW here are the links to the Bryston amplifier manuals again (the forum broke the links in yours, and I had to urlencode the square brackets):
Bryston 3B, 4B: http://old.bryston.com/PDF/Manuals/300006[3B4B].pdf
Bryston 7B: http://old.bryston.com/PDF/Manuals/300008[7B].pdf
I mean I can't really be talking about high hiss from the 12" speakers. But there's noise coming out, yes(sssss). I guess I can say the MF drivers have some hiss.
Sorry, what do you mean here? I don't know how feedback resistors relate to the noise value - or why they do that.
Yes and no. Don't forget that I'm doing this to set up a mastering rig. So I'll need to listen at high SPL every now and then, especially when it comes to dance music. With that sort of music, listening at high SPL is not painful in fact - because almost all the power goes to the bass.
The review you link to mentions very low output power. I wonder if it is enough to properly drive the inputs of the 10B?
Thanks a lot for the link. The noise floor doesn't keep on dropping though. Are you comparing the noise floor for 0 dBFS and -60 dBFS? If you look at the level vs THD+N graph, it seems like all the improvement you'll ever get is reached at -10 dBFS. So maybe it makes sense to always drive this unit at -10 dBFS and not more.
I was asking myself something like this. Should I be pulling it up to 13V rms? As John pointed out, op amp circuits will have the lowest THD+N somewhere close to, but noticeably below clipping. How to find out what that level is?
I did think of this. However, matching 8 channels would be a pain, and a stepped attenuator could be a pain due to clicks. Maybe a digitally controlled amplifier could work well here. But you really want your gain structure to be fixed and super reliable when doing mastering. So digital control would have to be very, very precise.
The noise just kinda falters sometimes. I haven't checked if it's from all the transducers, or just one of them. It's weird. I don't have signal level faltering when listening to music.
Yeah, I thought about doing something like that. I'll probably do this when replacing this analog crossover with a digital FIR one. I don't have a good ADC right now to be honest. But I don't think it's actually necessary to measure it... it's stated as Linkwitz-Riley, the cut off values are documented, the resistor values are designated as well. Shouldn't be an issue.
BTW here are the links to the Bryston amplifier manuals again (the forum broke the links in yours, and I had to urlencode the square brackets):
Bryston 3B, 4B: http://old.bryston.com/PDF/Manuals/300006[3B4B].pdf
Bryston 7B: http://old.bryston.com/PDF/Manuals/300008[7B].pdf
I mean I can't really be talking about high hiss from the 12" speakers. But there's noise coming out, yes(sssss). I guess I can say the MF drivers have some hiss.
Sorry, what do you mean here? I don't know how feedback resistors relate to the noise value - or why they do that.
Yes and no. Don't forget that I'm doing this to set up a mastering rig. So I'll need to listen at high SPL every now and then, especially when it comes to dance music. With that sort of music, listening at high SPL is not painful in fact - because almost all the power goes to the bass.
The review you link to mentions very low output power. I wonder if it is enough to properly drive the inputs of the 10B?
I don't really know. I'll have to figure out how to disassemble the thing and look at it, because the Bryston 10B schematic isn't exactly the same as the PMC 10B - there are subtle changes in configuration. I think I'll need a special set of square drivers, never needed that. Bryston use two different kinds of security screws on this thing... nice.
Thanks a lot for the link. The noise floor doesn't keep on dropping though. Are you comparing the noise floor for 0 dBFS and -60 dBFS? If you look at the level vs THD+N graph, it seems like all the improvement you'll ever get is reached at -10 dBFS. So maybe it makes sense to always drive this unit at -10 dBFS and not more.
I was asking myself something like this. Should I be pulling it up to 13V rms? As John pointed out, op amp circuits will have the lowest THD+N somewhere close to, but noticeably below clipping. How to find out what that level is?
What's the difference - low or high value resistors? I assume you are talking about feedback resistors, could you confirm please?
I did think of this. However, matching 8 channels would be a pain, and a stepped attenuator could be a pain due to clicks. Maybe a digitally controlled amplifier could work well here. But you really want your gain structure to be fixed and super reliable when doing mastering. So digital control would have to be very, very precise.
The noise just kinda falters sometimes. I haven't checked if it's from all the transducers, or just one of them. It's weird. I don't have signal level faltering when listening to music.
Yeah, I thought about doing something like that. I'll probably do this when replacing this analog crossover with a digital FIR one. I don't have a good ADC right now to be honest. But I don't think it's actually necessary to measure it... it's stated as Linkwitz-Riley, the cut off values are documented, the resistor values are designated as well. Shouldn't be an issue.
Last edited:
No problem 🙂
Oh, so you mean there are no distribution amplifiers?
Well, I don't know. There's that daughter board with two integrated op amps.. I don't really know what it does.
Here you go again 🙂
Bryston 10B schematics: Bryston: 10B SCHEMATICS : Free Download, Borrow, and Streaming : Internet Archive
BTW, thanks a lot again for pointing out that lower noise transistors could be used. I did think of cloning the boards with lower noise op amps, but I didn't know for a fact that you could also try to replace the transistors with lower noise ones.
Resistors have an inherent level of noise associated with them. This is called Johnson noise and it inserts itself as a noise source into a circuit when placed in certain locations.
Opamps also have small currents that flow through the input pins. These currents flow as a consequence of the devices operation and are intrinsic. If you place a resistor in series with an input pin this current still needs to flow so it flows through that resistor. The current flowing is random in nature and has the spectral content of a standard noise source. As the current is flowing through the input resistor it therefore induces a voltage proportional to the size of the resistor equal to ohms law V=IR. As the input current is fixed the bigger the resistor the bigger the voltage and the more noise you'll get.
Sometimes the application for an opamp will have a high series input resistance and thus will induce a high amount noise within this resistance if an opamp with a high input current is used.
Without knowing the specifications for the discrete opamps, used within the 10B, we can't make any assumptions as to the source of the 10Bs noise. Or rather if any particular stage is dominant in this situation.
Right but I highly doubt you'll need 700 watts driving a high sensitivity driver like the Volt radial. Especially as your speakers use a pair of them. They probably reach their linear excursion limits with far less power than this.
It's not very low it's just a tad lower than the standard. ~2.5dB lower.
Security screws are the worst. 😱
No of course it doesn't keep on dropping. The point was to highlight that calculations for how low noise a prospective preamp might have to be shouldn't just be done on the level of the noise floor with the DAC operating at 0dBfs.
You absolutely want to be running the DAC signal level at 0dBfs. If you were optimising the signal chain for playing a sinewave with the lowest distortion and highest SnR then dropping the signal level back a few dB, until you hit a sweet spot in performance, would be something you'd want to do. For music this isn't the case however. For music the overall system dynamic range is a much more important parameter, case in point with this system here. You've already got a limited dynamic range by the high noise of the active crossover and you're listening to it with constant hiss. To ensure maximum dynamic range from the DAC you need to use it without any digital attenuation.
It doesn't matter. Bryston have already told you it can handle 13VRMS without clipping. If you want to play it safe drop this back to 12.5VRMS. Yeah the distortion might rise a little but it will be inconsequential relative to the level of distortion the loudspeakers will be producing. If you were wanting to truly optimise the performance then this might be something you would consider at a later date but right now you've got a massive reduction in system THD+N because of the excessive amount of noise. Get that sorted first.
You'd need to use something like CS3318 multichannel volume control IC.
A reduction in noise could relate to a system instability that occasionally fixes itself but we shall overlook that for now because without an oscilloscope it will be difficult to diagnose.
Its stated as LWR for the standard 10B, but any half decent customisation for the device, by PMC, would use nonstandard electrical slopes to provide the correct acoustic slope after the loudspeakers roll off has been accounted for too.
There's no need to use FIR filters either. IIR are more than enough for speaker filtering. The filters inside the 10B are the analogue equivalent of IIR filters.
If you're going to be using a DSP for the crossovers you don't want to be using an ADC. You want to take the digital signal from the PC, run it through the DSP then use a multichannel DAC followed by a multichannel volume control. The signal chain for a low noise DSP based system is a pain to implement correctly too.
With regards to the 10B it's possible that one particular stage within it is causing most of the noise.
First of all I assume you've seen how noisy the speakers are with the power amplifiers powered on and their inputs shorted to ground?
Anyway I digress. You've tried the system with the input to the active crossover shorted to ground with no difference. Well you can short the various internal stages of the 10B to ground too. Once you have found out where the inputs to the sequential stages are you can short them to ground. Start with the stage nearest the outputs and work your way back to the inputs. It's possible that one stage is dominating the noise performance and modifying it, and it alone, could go a long way to solving your troubles here.
I'll try to make a schematic sometime this or next week.
First I'll need to fix my gain structure so that I am *able* to run them at full 700 watts. Then I'll see.
This may well be true for normal listening, but I believe in mastering you want to have low distortion more than you want infinite dynamic range. If you lower the signal to -10 dBFS on that chip then the noise drops by 4-6 dB but the distortion drops considerably more, so I think it's a good trade off. I've noticed myself backing off a good 10 dB as well because the mids become too harsh.
How do we know it can handle 13V rms? Where is the spec? I'm not sure where we got that from. I've read the brochure and the schematic again, but it doesn't mention anything like that I think.
I have some good tektronix oscilloscopes... just no place to put them right now!
Fairly certain that it's a standard 10B inside, just hard wired to be in 3-way mono mode.
Yeah I was just talking about ADC for measurement.
I haven't tried that specifically but I tried just unplugging the cable from the 10B's output and that quieted down the tweeter. So the amps are not noisy at all even with the input not shorted. I think that should show that it's not the amps, right?
Did you read this? A 1000 watt amp has a gain of 30db. So an automatic 30db rise in noise from whatever is before it. Then put that noise into a sensitive speaker. I think your wasting your time trying to make the cover less noisey.
Most power amplifiers have a gain around 30dB. Nothing wrong with that and nothing wrong with how powerful it is either. We aren't talking compression drivers here it's a standard soft dome tweeter.
Well I mean I'm certainly not going to build up the circuit for the discrete opamp in spice and simulate it for its performance. I don't see the point in drawing up a schematic for the opamps themselves.
Where are you controlling the volume from? The Fiio's volume control is analogue and uses a digitally controlled potentiometer it doesn't adjust the digital signal level at all.
Because it says so in the specification sheet johnmath linked to before.😕 Vout max.
Well if it's oscillating that would explain the changes in noise level.
If that's the case then you can probably design a much better crossover. Textbook LWR filters are rarely optimal even with active designs.
Lots of PC soundcards have a line-in. Or laptops.
Then it is the noisy active crossover.
Try plugging the output of the Fiio directly into the poweramp to the tweeter and see how noisy that is. Then we can get a feel for how quiet the system could potentially be.
5th element, thank you for taking over the discussion.
The problem with the current setup should be obvious - the 10B crossover still has ~28 or 22 dB of unused headroom (depending on the amplifier input sensitivity setting) when the amplifier hits 0 dB of headroom AKA clipping. So there's an easy 20 dB reduction in noise and distortion to be had simply by putting in the missing 20 dB pads at the amplifier inputs. That is enormously more improvement than the couple of dB that might be found in the crossover circuitry. For Cheater that is his problem solved. Q.E.D.
If necessary to achieve the SPLs needed, some low noise gain would be added at the output of the Fiio. It shouldn't be hard to keep the noise floor of that below the noise floor of the Fiio.
I am retired and out of touch of who makes what. I can't find a unbalanced to balanced consumer audio interface with decent specs. Radio Design Labs has the best one I've seen but I was looking for something with more dynamic range / lower noise floor. Is there no demand for this type of product anymore?
RU-AFC2 ‐ Stereo Audio Format Converter
The problem with the current setup should be obvious - the 10B crossover still has ~28 or 22 dB of unused headroom (depending on the amplifier input sensitivity setting) when the amplifier hits 0 dB of headroom AKA clipping. So there's an easy 20 dB reduction in noise and distortion to be had simply by putting in the missing 20 dB pads at the amplifier inputs. That is enormously more improvement than the couple of dB that might be found in the crossover circuitry. For Cheater that is his problem solved. Q.E.D.
If necessary to achieve the SPLs needed, some low noise gain would be added at the output of the Fiio. It shouldn't be hard to keep the noise floor of that below the noise floor of the Fiio.
I am retired and out of touch of who makes what. I can't find a unbalanced to balanced consumer audio interface with decent specs. Radio Design Labs has the best one I've seen but I was looking for something with more dynamic range / lower noise floor. Is there no demand for this type of product anymore?
RU-AFC2 ‐ Stereo Audio Format Converter
Last edited:
The Fiio has both balanced and unbalanced outputs anyway so I'm not sure why one of those would be needed here?
Not strictly balanced, at least not in the professional audio sense, the Fiio outputs are push-pull AKA bridged outputs.
In any case it is clear that some gain is needed on the output of the Fiio. Gain is almost always implemented in unbalanced circuitry. To used the balanced output implies a lot more circuity. The first chain is one way, the second chain is with a consumer audio interface, which eliminates 3 stages that can introduce N+D:
1.DAC out =unbal ➔ 2.driver stage (line out) =unbal ➔ 3.headphone out =unbal ➔ 4.bridge out =psudo bal ➔ 5.balanced input stage =bal ➔ 6.preamp circuit =unbal ➔ 7.balanced line driver =bal ➔ etc.
1.DAC out =unbal ➔ 2.driver stage (line out) =unbal ➔ 3.preamp circuit =unbal ➔ 4.balanced line driver =bal ➔ etc.
(Note I am making some assumption about the circuitry in the Fiio; I'd be surprised if they are wrong though.
In any case it is clear that some gain is needed on the output of the Fiio. Gain is almost always implemented in unbalanced circuitry. To used the balanced output implies a lot more circuity. The first chain is one way, the second chain is with a consumer audio interface, which eliminates 3 stages that can introduce N+D:
1.DAC out =unbal ➔ 2.driver stage (line out) =unbal ➔ 3.headphone out =unbal ➔ 4.bridge out =psudo bal ➔ 5.balanced input stage =bal ➔ 6.preamp circuit =unbal ➔ 7.balanced line driver =bal ➔ etc.
1.DAC out =unbal ➔ 2.driver stage (line out) =unbal ➔ 3.preamp circuit =unbal ➔ 4.balanced line driver =bal ➔ etc.
(Note I am making some assumption about the circuitry in the Fiio; I'd be surprised if they are wrong though.
What does "bridged" mean? I've never heard of that when talking about DACs... would you mind explaining please?
I can't use unbalanced from the DAC to the hi fi. It results in a nasty ground loop. I had to cut pin 1 on cables going from the DAC to the xo - both the single ended output and the balanced output cables. And whatever amplifier / etc I'll be using will be powered off the same strip as the Bryston electronics, so I'll need balanced signalling.
I've found that the "at least 13V rms output Vmax" is for unbalanced output - the balanced output model has at least 24V rms output Vmax.
BTW,
I've finally found a picture of the Bryston 10B LR. I'm now fairly certain that that is what mine is inside. The Bryston 10B LR has only one set-rotary-switch (knobless) per side (two total), while the normal version (that's apparently not Linkwitz-Riley) has three knobs per side. The PMC 10B also has only one rotary switch, but the front panel was made thicker so the switch wouldn't make it through.. you have to take the front panel off for that.
Also it occurred to me that in the manual it says that the three-way mono mode is only available in the LR model, and that all filter slopes are 24 dB/oct, but they didn't mention if that's /only/ in mono three-way mode, or if it's also like this in stereo mode, but limited to the LR model.
The attachments show:
1. Bryston 10B-LR crossover (that's the most like my PMC 10B)
2. Bryston 10B-SUB (basically a 10B-STD but with lower cutoff values for use with subwoofers)
3. Bryston 10B-STD (the usual one for two-way amplification)
4. PMC 10B
BTW, in the first one, the model designation is 10B-LRB2. The LR is for Linkwitz-Riley, B is for Balanced, and 2 is for the generation, as it belongs to the SST2 series. The PMC electronics I have are SST2, too.
Attachments
Last edited:
It's true that a better gain structure will definitely help with this, and maybe even make it not a problem any more. We're going to figure out with time what the best gain settings will be and what the best preamps will be. However, figuring out the 10B and why it's noisier than thought is something I'm very curious about and it's very fun to try and figure this stuff out.
Bridging is when two channels of amplification are used to double output voltage swing. It is reasonably common to see a power amplifier with a 'bridge' switch on the back panel. One channel is inverted so its output swings negative for a positive input. Then the load is connected to the output terminals of the two amplifiers in push - pull arrangement, and since both channels are use the amplifier is now mono only. The ground terminal is not used for connecting to the speakers. The reason it isn't strictly a "balanced" output is just that a true balanced output has to suppress common mode signal at least -60dB or hopefully a lot more, and this is unlikely to be achievable with this type of headphone amplifier push/pull output. Nevertheless, it makes perfect sense to use it as you have.
Bryston's spec for balanced output is =29dBv = 28Vac. 13Vac = 22dBv which is -6dB from balanced, as expected.
With -100dBv noise specification and +29dBv before, giving 129dB dynamic range, I do not understand why you think the 10B is noisy. The Bryston *B amplifiers' noise is 110dB below clipping i.e. 19dB worse than the 10B, and the Fiio -115dB i.e. 14dB worse than the 10B (but that's optimistic because it is a weighted measurement). The 10B has a much better dynamic range that the other two items in your rig, but because of your gain structure you are just taking advantage of the 10B's inherently superior dynamic range. The 10B is the quietest component you have if used properly.
If these specifications are in fact correct dropping the output of the 10B with 20dB pads will lower it's noise to far below that of the amplifiers' and still drive the amplifiers to rated power, which requires +5dBv.
You need to fix the mismatch between the Fiio's balanced output ~10dBv which is 19dB below the optimal drive level for the 10B's balanced inputs.
I have been assuming balanced in and out connections on the 10B, but you unit is unbalanced? That does not change its dynamic range of 129dB, but does change the optimum dBv in and out to be -6dB compared to balanced, so a 14dB pad for the amplifier (or 8dB for low gain mode).
BTW your Brystons have a bridge switch on the rear.
My 10B and all the amplifiers are fully balanced I/O.
I think you're right about the dynamic range of the 10B.
I will need to find a good +20dB preamp circuit.
What sort of parameters should I be looking for?
I guess it should have at least around 125 dB dynamic range and put out 26 V rms into balanced. Low distortion. What else should I be on the look out for?
I think you're right about the dynamic range of the 10B.
I will need to find a good +20dB preamp circuit.
What sort of parameters should I be looking for?
I guess it should have at least around 125 dB dynamic range and put out 26 V rms into balanced. Low distortion. What else should I be on the look out for?
I can't seem to find an off the shelf line amplifier product that has THD+N specs as good as the 10B, so you may need to make one yourself. Noiseless gain is easy enough if you use a transformer as an interface, but then you won't get much better than 0.01% THD. The distortion will be mostly 3rd order, so it will impart a itsy weeny bit of valve-ish sound.
You have a low source impedance for an active design to get equivalent input noise low enough. Getting op amp circuitry to work as well as it can is a bit of an art and not something I know well enough to design competently.
There are balanced in and out op amps like the OPA 1632 you could use for the input stage with 20dB of gain and follow with a discreet output drive, perhaps copy the Bryston circuit which operates in push-pull mode to get double voltage swing from 24 volt rails, with 6dB of gain to give you 26dB gain total. The OPA1632 differential outputs could each drive one discrete op amp so only two Bryston type active stages required for each channel, and ±24 volts for the output stages and ±15 volts for the input stage.
Hopefully some who knows more than me can either reference an off the shelf unit, or a decent DIY design. The best off the shelf unit I can find, whilst being better than the Fiio still has less dynamic range than the 10B.
You have a low source impedance for an active design to get equivalent input noise low enough. Getting op amp circuitry to work as well as it can is a bit of an art and not something I know well enough to design competently.
There are balanced in and out op amps like the OPA 1632 you could use for the input stage with 20dB of gain and follow with a discreet output drive, perhaps copy the Bryston circuit which operates in push-pull mode to get double voltage swing from 24 volt rails, with 6dB of gain to give you 26dB gain total. The OPA1632 differential outputs could each drive one discrete op amp so only two Bryston type active stages required for each channel, and ±24 volts for the output stages and ±15 volts for the input stage.
Hopefully some who knows more than me can either reference an off the shelf unit, or a decent DIY design. The best off the shelf unit I can find, whilst being better than the Fiio still has less dynamic range than the 10B.
There is this line amplifier that exceeds the required specifications, but it is not clear if it can proved 20dB of gain:
Benchmark LA4 Line Amplifier - Benchmark Media Systems
I don't remember if I linked to this article before; it is very detailed and in lines up with my understandings, although sometimes expressed in different ways:
Balanced vs. Unbalanced Analog Interfaces - Benchmark Media Systems
Benchmark LA4 Line Amplifier - Benchmark Media Systems
I don't remember if I linked to this article before; it is very detailed and in lines up with my understandings, although sometimes expressed in different ways:
Balanced vs. Unbalanced Analog Interfaces - Benchmark Media Systems
I've run into an issue. In order to better disassemble the 10B, I need to undo a 1mm square screw. That's 1mm at the base of the screw, not the diagonal. Short of manufacturing one on my own, I don't know what to do, I can't find one anywhere. The tip is supposed to be SQ00 or S00. But I can only find down to SQ0...
I am located in Austria, but honestly I can't find those /anywhere/. The few I could find were $100 for a set of like 5, and that's just ridiculous...
I am located in Austria, but honestly I can't find those /anywhere/. The few I could find were $100 for a set of like 5, and that's just ridiculous...
Last edited: