I assumed nothing.
Hi George, back from Europe, nice to escape some of the winter here.
I don't see any bottle neck as such. The current on Pin 5 is fixed.
What is the value on Pin 5 here? If much lower than 100K, then they are in parallel and become the final Pin 5 value. The output here is current and hence very high Z indeed. Technically I still prefer a buffer, but if the Pin 5 total value is zero Ohm, then no output and yet the output Z would also be zero (a thought experiment only) and thus if the total Pin 5 value is kept small, then likewise the Z. It then comes down to whether you have enough gain (and possibly headroom limitations?). Allowing for these, then if he likes the sound, then I should not be critical.
Point I am trying to make: The final value on Pin 5 sees current being pushed through it. Current drive (steering?) is something that can be difficult to wrap one's mind around as we find it much easier and intuitive to think in terms of voltage. The total current available on Pin 5 is actually fixed (for a particular level) and can be calculated. With a buffer, the current is a function of the load, but on Pin 5 it is not. The load makes no difference. Sounds weird, but true.
Cheers, Joe
.
Joe watch the TZ on the scope with <100kohm load and cd full level test 1khz or 10khz signal sine wave, you will see the output drop compared to a much higher value say 1megohm. Thus making it a voltage divider.
As the TZ has very weak output, to my ears it causes a bottleneck to the dynamic capabilities.
This is also why Charles Hansen used a second high input impedance (7megohm) buffer after the TZ output of the I/V stage in his Ayre CX-7xe dac and CX-7e cdp where he uses 8 x AD844's. I'd show the the circuit, but I'd have to the kill you.
Cheers George
As the TZ has very weak output, to my ears it causes a bottleneck to the dynamic capabilities.
This is also why Charles Hansen used a second high input impedance (7megohm) buffer after the TZ output of the I/V stage in his Ayre CX-7xe dac and CX-7e cdp where he uses 8 x AD844's. I'd show the the circuit, but I'd have to the kill you.
Cheers George
Last edited:
Of course it is, because current is what comes out of Pin 5 and voltage comes out of Pin 6. So the apparent load sensitivity on Pin 5 is not a sign of it being weak. The real load in the R value on Pin 5 to ground and not the remote load which is much higher. The drop is not a result of weakness but rather that the additional Pin 5 load already is in parallel with the added remote load. Of course the remote load has an effect, it is supposed to. The current is not restricted and remains the same, so there is no bottleneck in that sense.
And parallel resistors here do not become voltage dividers, but actually become current dividers. And Charlie Hansen perfectly understands this - and I know he takes the output of Pin 5 and it goes through a buffer of his choice/design, I would have done likewise. I assure you that you don't need to do any killing in this instance.
I can understand the confusion, that is the nature of the actual output of the transconductance which is on Pin 5 and not on Pin 6. One is meant to sag voltage wise and the other is not. It gets even weirder when you consider that a voltage output can produce current that deviates from zero phase angle, but a current output has no current phase shift because it can only produce a zero phase angle. This I believe is one pointer to why transconductance circuits like this can have extra-ordinary bandwidths, which the AD844 has. No matter what the load on Pin 5, there is no phase shift into the buffer. But it will sag voltage wise, that is the nature of the beast.
Having said that, for commercial reasons you can never take the output from a current output, so you have to use a buffer. And I agree, the internal buffer of the AD844 is no slouch and more than respectable, if not absolutely the best. And it gets better with a constant current into a shunt regulator. Without feedback, the buffer does produce noticeable uncorrected distortion, but I would rather have that than artificially lowering it.
Cheers, Joe
.
Last edited:
Of course this is why the Ayre has a very high input impedance buffer after the TZ I/V stage, and what I found out as well, this is why one should use higher than optimationman wanted to use, 250kohm + resistor that goes to ground after it before the input transistor or tube, whatever he uses, as this loads the TZ down even more.
Cheers George
Cheers George
And Joe, everything I have put forth wether Lightspeed, AD-844 i/v stages, mods to many dac/cdp's and many other projects, I have posted circuit diagrams on how to do it for the members. As is they that will build them as it is diy.
So if you want to say you have a better way please post up circuit diagram of your "better way" so the members can build and critique it here on the forums.
Cheers George
So if you want to say you have a better way please post up circuit diagram of your "better way" so the members can build and critique it here on the forums.
Cheers George
I don't know where that came from?
Seems out of context my the posts I have made. That was not the topic.
.
Joe your telling me and others to do it another way, so please so they (the members) can do it and appraise your way.
Post up the AD844 say with the TDA1541 or PCM1704 with the circuit you suggest "And it gets better with a constant current into a shunt regulator." So then they can try it and post opinions.
Cheers George
Post up the AD844 say with the TDA1541 or PCM1704 with the circuit you suggest "And it gets better with a constant current into a shunt regulator." So then they can try it and post opinions.
Cheers George
No, George, I was not. Please read my posts again. I was mentioning pros and also cons, as well as analysing the output on Pin 5 which is not voltage and thus behaves in a non-standard manner, which you yourself have found out in a round-about way. There are no voltage dividers when the output is current, the current gets split into two parallel resistors that equates to a single value. I was not promoting any schematic and there was no need to in the context of my posts today.
Think of Pin 5 as a current send output. That current has to go somewhere, analyse that and you will see what I mean.
Cheers, Joe
.
I have no wish to do it Joe, I say show people, how it's done, as you have been asked many times before with other mods on other threads to other things to show what you mean.
This way other members can see what you mean and so do it for themselves, not to purchase ready made units from halfway around the world, and pay $>100 + postage 10 x + what it should cost to diy.
Cheers George
Last edited:
Huh? George, you are not making any sense to me? I am not selling anything here, I am not proposing any circuits, I find your responses to be curious...
I have posted heaps of schematics, here on diyaudio and elsewhere, but what has that got anything to do with my posts today... nothing!
.
Joe you've been here before on other threads, just suck it up. And show things (in circuit diagrams) that you suggest members to do.
As this one is just plain dumb, for a $2 I/V AD844 stage you want members to spend over $100 for aready made shunt regultaor for it.
When they could build it for less than $5 if you showed them the circuit, or are you an agent for this ready made shunt regulator?
Cheers George
As this one is just plain dumb, for a $2 I/V AD844 stage you want members to spend over $100 for aready made shunt regultaor for it.
When they could build it for less than $5 if you showed them the circuit, or are you an agent for this ready made shunt regulator?
Cheers George
back to the AD844
The output of pin 5 goes to a 250k pot that acts as a volume control. The wiper goes directly to the grid of the input stage 12ax7a. I found that the result sounds much better that the buffer in the AD844. I have tried both single and dual buffer connections but I like the sound of the direct connection best. It does result in the description that George has said resulted in connecting up the buf3 he uses. How much more would be gain in sound is still in question. I think if there were to be an increase in dynamics etc. it would not result in a natural presentation. There is just a realistic sound as is, just keep turning up the volume and just the loudness increases without new artifacts or strain.
I will have time today to run a freq sweep just to see what really is coming out. Smooth it is however. I will be installing different resistors to pin 5 as I do have plenty of gain to work with. Fun stuff when sound improves, at least to my ears.........
The output of pin 5 goes to a 250k pot that acts as a volume control. The wiper goes directly to the grid of the input stage 12ax7a. I found that the result sounds much better that the buffer in the AD844. I have tried both single and dual buffer connections but I like the sound of the direct connection best. It does result in the description that George has said resulted in connecting up the buf3 he uses. How much more would be gain in sound is still in question. I think if there were to be an increase in dynamics etc. it would not result in a natural presentation. There is just a realistic sound as is, just keep turning up the volume and just the loudness increases without new artifacts or strain.
I will have time today to run a freq sweep just to see what really is coming out. Smooth it is however. I will be installing different resistors to pin 5 as I do have plenty of gain to work with. Fun stuff when sound improves, at least to my ears.........
optimationman, seeing that the only load the TZ sees is the 250kohm pot then this should be ok as the slight drop of level on the TZ I found was at 100kohm.
I found direct connection from TZ to my amps using the digital domain VC in my cdp was very very nice to listen to but it was almost too polite, it reminded me of the difference in redbook replay between a R2R Multibit dac and a Delta Sigma/ess dac, when done right multibit grips your attention, and SD well I can almost go to sleep to.
One other point going from the TZ output, as it is very high in output impedance watchout for interconnect cable capacitance to the amp it could roll off the highs.
Cheers George
I found direct connection from TZ to my amps using the digital domain VC in my cdp was very very nice to listen to but it was almost too polite, it reminded me of the difference in redbook replay between a R2R Multibit dac and a Delta Sigma/ess dac, when done right multibit grips your attention, and SD well I can almost go to sleep to.
One other point going from the TZ output, as it is very high in output impedance watchout for interconnect cable capacitance to the amp it could roll off the highs.
Cheers George
sound is great
George, thanks for the heads up. I find the sound very lively and clear. Much better than the in chip buffer. Not sure why but the sound is smoother class a vs class b? I could still hear I guess for the lack of another work, artifacts in very familiar recordings and those artifacts are now gone. I am using emitter/follower dedicated to the dacs and one for the AD844, with optimal loads. The decoupling on the AD is .01/4.5 uf ceramics with a 150 uf silmic. These are connected to the pin and the ground plane right under the chip. Not sure if this is optimal. I didn't/haven't done the large ceramic +- buss mod yet, and that may have improved the in chip buffer, ?? I have the nulling circuit that you used, but would it be advisable to go with the AD844 null system if we are using only the TZ output?
The interconnects are 2 ft of Beden 1695a which is only 32pf, and there is no RCA jacks as they are soldered to the termination thru a Riken 150 ohm resistor.
I originally had 2.8k/450pf on TZ and just tried 1.5k/1n. Couldn't hear much change. How low can the resistor go. I was thinking of trying something in the 1.0k range as I still have some of gain to play with.
I haven't listened to to many Delta Sigma players, but have modded several TDA1541 machines and I keep this one.
George, thanks for the heads up. I find the sound very lively and clear. Much better than the in chip buffer. Not sure why but the sound is smoother class a vs class b? I could still hear I guess for the lack of another work, artifacts in very familiar recordings and those artifacts are now gone. I am using emitter/follower dedicated to the dacs and one for the AD844, with optimal loads. The decoupling on the AD is .01/4.5 uf ceramics with a 150 uf silmic. These are connected to the pin and the ground plane right under the chip. Not sure if this is optimal. I didn't/haven't done the large ceramic +- buss mod yet, and that may have improved the in chip buffer, ?? I have the nulling circuit that you used, but would it be advisable to go with the AD844 null system if we are using only the TZ output?
The interconnects are 2 ft of Beden 1695a which is only 32pf, and there is no RCA jacks as they are soldered to the termination thru a Riken 150 ohm resistor.
I originally had 2.8k/450pf on TZ and just tried 1.5k/1n. Couldn't hear much change. How low can the resistor go. I was thinking of trying something in the 1.0k range as I still have some of gain to play with.
I haven't listened to to many Delta Sigma players, but have modded several TDA1541 machines and I keep this one.
hello, what input impedance this buffer?
http://www.diyaudio.com/forums/solid-state/225255-jlh-buffer-homage-john-linsley-hood.html
i want use ad844 with thoses.
http://www.diyaudio.com/forums/solid-state/225255-jlh-buffer-homage-john-linsley-hood.html
i want use ad844 with thoses.
I have not used Abrxilito's dac offset nulling circuit for a while now as it is a very sensitive area to have things attached to it. I use pins 1 & 8 of the 844 if using it's buffer, or pins 1 & 5 of the BUF03 when using it.
As for the TZ resistor, I get as much gain from it as I need to get 2v with a larger resistor and a smaller cap for the low pass, as maybe a larger could cause instability in the I/V stage.
Cheers George
As for the TZ resistor, I get as much gain from it as I need to get 2v with a larger resistor and a smaller cap for the low pass, as maybe a larger could cause instability in the I/V stage.
Cheers George
You would have to ask Barrie Gilbert the designer, as only he has the full schematic of what inside, the data sheet only gives a Figure 31. Simplified Schematic.
I taked to him a while back, and was very proud of the AD844, and he was very interested in what we were doing with it esspecially the stacking and no feed back use, he did say he'd look through all his archived circuit diagrams files and send it to me, but he never did. I'd say he is very busy being contracted out to develop for big companies.
Cheers George