DAC AK4499EX Project
- Thread starter piglet11
- Start date
Has anyone heard from piglet11 recently? I had been exchanging messages with him about the PCB for this project, then nothing...
As I havent heard from Piglet about this project in some time, I pulled the trigger on a Chinese kit from GeeDIY (partially assembled and minus the 4499/4191). I am in the process of modifying the board and repopulating some areas with better parts. In a number of areas this kit is not as well though out as the module Piglet has shown here appears to be. I am working on it to make it better.
As this thread is probably read by some people rolling their own 4499EX, I thought I might find someone who has come across this already.
The AK4497EX "datasheet" says, on page 29 that the I/V opamps must be powered on after the 4499 and powered off before the 4499. This it doesnt appear to have any control in place to do this. The +/-15VDC for the LP filter and I/V converter cone in directly to both functions so their power on/off relative to the power for the 4499 seems to be completely dependent on how quickly the 15V comes up after power is applied compared to the three 7-9V supplies for the 4499 board. There is nothing in the GeeDIY online instructions for the 4499 module that suggest this is needed, and nothing on the board that suggests it is addressed in any way on the board. The AK4499 datasheet DOES state that "There is a possibility of IC destruction due to breakdown of the withstanding voltage of the analog output pins (IOUTLP/LN/RP/RN) if the power supply of the external operational amplifier is turned on before power up the AK4499EX. Therefore, connect a Zener diode (VRWM = 6 to 7 V) between each VDDL/R and VSSL/R if the power up/down sequence shown in Figure 13 cannot be followed." I believe these zeners are shown on the drawing on page 45 of the datasheet - labeled ZD, and on the AK4499EX evaluation board manual schematic page 45 labeled D1 and D2.
Question. Has anyone implemented this zener arrangement in lieu the required of power supply sequencing when using the 4499EX? From the description by AKM it looks as though limiting the voltage on VDD to under 6-7V will somehow protect damage to IOUT in the event of a damaging voltage being fed back from the I/V opamps during power up and down.
As this thread is probably read by some people rolling their own 4499EX, I thought I might find someone who has come across this already.
The AK4497EX "datasheet" says, on page 29 that the I/V opamps must be powered on after the 4499 and powered off before the 4499. This it doesnt appear to have any control in place to do this. The +/-15VDC for the LP filter and I/V converter cone in directly to both functions so their power on/off relative to the power for the 4499 seems to be completely dependent on how quickly the 15V comes up after power is applied compared to the three 7-9V supplies for the 4499 board. There is nothing in the GeeDIY online instructions for the 4499 module that suggest this is needed, and nothing on the board that suggests it is addressed in any way on the board. The AK4499 datasheet DOES state that "There is a possibility of IC destruction due to breakdown of the withstanding voltage of the analog output pins (IOUTLP/LN/RP/RN) if the power supply of the external operational amplifier is turned on before power up the AK4499EX. Therefore, connect a Zener diode (VRWM = 6 to 7 V) between each VDDL/R and VSSL/R if the power up/down sequence shown in Figure 13 cannot be followed." I believe these zeners are shown on the drawing on page 45 of the datasheet - labeled ZD, and on the AK4499EX evaluation board manual schematic page 45 labeled D1 and D2.
Question. Has anyone implemented this zener arrangement in lieu the required of power supply sequencing when using the 4499EX? From the description by AKM it looks as though limiting the voltage on VDD to under 6-7V will somehow protect damage to IOUT in the event of a damaging voltage being fed back from the I/V opamps during power up and down.
So long as the I/V opamps are powered on/off at the same time as the rest of the dac there doesn't seem to be a problem. For example, AKM didn't use the zeners on the AK4499 evaluation board. They just powered the opamps off of the same +-15v that powered the rest of the dac board.
However, what I did was sequence the turn on time of the +5v power to the eval board so it came on first. Then I used big filter caps for that power supply so that if there was a power failure, the +-15v power supply for the opamps would decay to zero much faster than the +5v power would decay. That seemed to make it safe; never had a failure using that scheme.
However, what I did was sequence the turn on time of the +5v power to the eval board so it came on first. Then I used big filter caps for that power supply so that if there was a power failure, the +-15v power supply for the opamps would decay to zero much faster than the +5v power would decay. That seemed to make it safe; never had a failure using that scheme.
Thanks for that info. I am entertaining the notion of having the 9V I supply to this board, gate the output of the +/-15V supply, it already gates the 3.3V power to the USB board and Arduino display.
You say AKM didnt use the zeners, yet they are in the schematics I received from them for their evaluation board as I indicated above, page 45, D1, D2.
You say AKM didnt use the zeners, yet they are in the schematics I received from them for their evaluation board as I indicated above, page 45, D1, D2.
IIRC, the AK4499EQ (not AK4499EXEQ) evaluation board AKM sent me showed protection zeners on the schematic, but when I went looking for them on the physical board they didn't exist. Again IIRC, there were however some Schottky diodes to protect against reverse bias somewhere around there too, which did exist. Not too surprising they left the zeners off since AKM did warn that using them could affect the sound.
All that having been said, I stopped working with IC dac chip based dac designs some time after AK4499EQ. Found that diy discrete resistor DSD dacs can sound subjectively far better. Not necessarily lower-cost to build though.
All that having been said, I stopped working with IC dac chip based dac designs some time after AK4499EQ. Found that diy discrete resistor DSD dacs can sound subjectively far better. Not necessarily lower-cost to build though.
Last edited:
Do you mean filter caps? If so they will reduce the peak voltage of a spike above the regulated voltage, (5V in the case of VDD) but they arent big enough do that for more than a few milliseconds unless the source of the additional voltage has a pretty high impedance. If a higher voltage remains present, they will charge to the higher voltage pretty quickly.
The source of the potentially destructive voltage must be from the I/V opamp outputs. I can visualize conditions wherein the opamp output jumps to near either the plus or - voltage rails if already power on while the DAC is booting up and turning on its Voffset and OPIN outputs if they dont come up at exactly the same time.
What I dont understand is exactly what the electrical relationship is between the OPIN outputs and VDD, where a zener would help avert damage.
The source of the potentially destructive voltage must be from the I/V opamp outputs. I can visualize conditions wherein the opamp output jumps to near either the plus or - voltage rails if already power on while the DAC is booting up and turning on its Voffset and OPIN outputs if they dont come up at exactly the same time.
What I dont understand is exactly what the electrical relationship is between the OPIN outputs and VDD, where a zener would help avert damage.
I managed to get my signal names backwards. That changes what I said some. There is no DC path from the opamp output back to OPIN in the recommended I/V layout. There is a small feedback cap in that path so only a very narrow spike could get from the opamp output to OPIN. There are DC paths from the opamp - input to OPIN and the opamp output to IOUT, IOUT the stated problem, not OPIN. So, everything above in message #130, where I referred to OPIN, should have said IOUT.
IIUC, at least for AK4499EQ the internal breakdown voltage of the dac chip when it is unpowered is around 7-8v, and the zener is to keep it from getting that high. Something like that.
...Which is why I kept the +5v and +3.3v power on longer than the opamp power. At least the dac chip was biased up to 5v. Also ran the opamps on +-11v which is what Topping used in the original D90. I thought it sounded best at that voltage anyway, although Topping may have been motivated to use those rail voltages by trying to protect the analog switch IC they used for output muting. Bottom line for me was that meant the worst case opamp outputs could only go to about +6v over the biased-on level of the dac chip under those conditions.
One thing I found out roughly around that time is how much of a problem using a shared ground for the +-11v rails could be. When I kept those voltage regulators fully isolated from each other and only grounded each one at the load, the sound clearly got better. IOW, the effects of common-impedance-coupling on a shared ground return wire can be quite audible.
Last edited:
One thing I found out roughly around that time is how much of a problem using a shared ground for the +-11v rails could be. When I kept those voltage regulators fully isolated from each other and only grounded each one at the load, the sound clearly got better. IOW, the effects of common-impedance-coupling on a shared ground return wire can be quite audible.
(Put your comment in italics because for some reason the forum isnt giving me a "Quote" button for the current last entry in the thread)
Interesting. I take it the +/- regulators didnt share a common ground at the regulator?
(Put your comment in italics because for some reason the forum isnt giving me a "Quote" button for the current last entry in the thread)
Interesting. I take it the +/- regulators didnt share a common ground at the regulator?
Correct. Now I only use mutually isolated positive regulators for everything. If I want a negative rail, I just connect the + regulator output terminal to ground at the load, and the - terminal of the regulator to the negative rail. For this to work each regulator has its own isolated power transformer secondary winding (no center taps).
BTW, there is no quote button for the last post in a thread. However, if you highlight text of the post, a button will appear to let you quote it.
Attachments
Last edited:
Still mulling over this caution regarding powering on the +/-15V before or off after the VDD is on/off.
One solution I have come up with is this...
I use one of the Chinese implemented Jung/Didden super regulators for the +/-Analog DC with a pre-regulator feeding each. I went looking for an easy way to enable/disable it. I have the control logic (0V-5V) in place to do something like this if the regulators can be shut down and started up from a TTL source. One thing I am avoiding is actually powering up and down the regulators since the time from applying or removing power to actually having it up or down is fairly long compared to the9V feeding the supplies on the DAC board itself.
I found that I can bring the regulators down to 5V quickly by jumpering the resistor (1k5 in this case) with something around 200 ohms from the regulator output to the opamp input. Doesnt seeem to hurt anything in the regulator and the DAC opamps I am using are minimally functional at +/-5V so nothing odd should happen there. When powered on initially with the jumpered resistor in place the supplies quickly ramp up to 5V, then when I switch out the resistor they step up to +/-15V (I have decided to reduce the supply rails to 12V for reasons associated with the opamps, but that is another matter). Upon AC power removal, my control circuit instantly switched in the resistor again and that pulls the rails back down to 5V slightly before the 9V feeding the DAC 5V, 3.3V and 1.2V regulators decays to around 5V then with decaying AC input the rails go on down to zero.
Short of having a designed in shutdown, I kind of like this as it has no effect on the regulator outputs when they are operating normally (resistors switched out) and it doesnt add any potentially performance quality degrading devices in series with the voltage rails themselves.
I am assuming that +5V on the analog opamp + rail isnt going to be "a possibility of IC destruction" situation, given the recommended solution is to use 6-7V zeners across the VDD/VSS rails. I will put something like Schottky diodes from VDD to VSS to prevent significant negative excursions in case -5V is a problem (the data sheet says -.3V is about the max permitted on any input pin). A reverse biased Schottky diode shouldnt create any unwanted noise on VDD...Unlike a zener 6-7V with a 5V level on it which is probably just starting into zener breakdown and creating some noise as a result. I might be wrong about the +/-5V rails being safe, since it isnt clear to me exactly what leads to the destruction problem within the 4499 chip.
By the way, I tested a few 6.8V zener diodes I have and found the current in the reverse bias direction is a few microamps up to about 5.5, where starts to increase dramatically. From 4V to 5.5V the current does creep up slightly but it is right at the lower limit of my meter so telling exactly how much is sort of a guess. Also the diodes I have are NOS 1n4736 from Fairchild. What I saw may not be typical of newer On-semi, or imported equivilants.
One solution I have come up with is this...
I use one of the Chinese implemented Jung/Didden super regulators for the +/-Analog DC with a pre-regulator feeding each. I went looking for an easy way to enable/disable it. I have the control logic (0V-5V) in place to do something like this if the regulators can be shut down and started up from a TTL source. One thing I am avoiding is actually powering up and down the regulators since the time from applying or removing power to actually having it up or down is fairly long compared to the9V feeding the supplies on the DAC board itself.
I found that I can bring the regulators down to 5V quickly by jumpering the resistor (1k5 in this case) with something around 200 ohms from the regulator output to the opamp input. Doesnt seeem to hurt anything in the regulator and the DAC opamps I am using are minimally functional at +/-5V so nothing odd should happen there. When powered on initially with the jumpered resistor in place the supplies quickly ramp up to 5V, then when I switch out the resistor they step up to +/-15V (I have decided to reduce the supply rails to 12V for reasons associated with the opamps, but that is another matter). Upon AC power removal, my control circuit instantly switched in the resistor again and that pulls the rails back down to 5V slightly before the 9V feeding the DAC 5V, 3.3V and 1.2V regulators decays to around 5V then with decaying AC input the rails go on down to zero.
Short of having a designed in shutdown, I kind of like this as it has no effect on the regulator outputs when they are operating normally (resistors switched out) and it doesnt add any potentially performance quality degrading devices in series with the voltage rails themselves.
I am assuming that +5V on the analog opamp + rail isnt going to be "a possibility of IC destruction" situation, given the recommended solution is to use 6-7V zeners across the VDD/VSS rails. I will put something like Schottky diodes from VDD to VSS to prevent significant negative excursions in case -5V is a problem (the data sheet says -.3V is about the max permitted on any input pin). A reverse biased Schottky diode shouldnt create any unwanted noise on VDD...Unlike a zener 6-7V with a 5V level on it which is probably just starting into zener breakdown and creating some noise as a result. I might be wrong about the +/-5V rails being safe, since it isnt clear to me exactly what leads to the destruction problem within the 4499 chip.
By the way, I tested a few 6.8V zener diodes I have and found the current in the reverse bias direction is a few microamps up to about 5.5, where starts to increase dramatically. From 4V to 5.5V the current does creep up slightly but it is right at the lower limit of my meter so telling exactly how much is sort of a guess. Also the diodes I have are NOS 1n4736 from Fairchild. What I saw may not be typical of newer On-semi, or imported equivilants.
I think it should be designed to ride out a power failure, then restoration of power by the power company without any problems. Fast shutdown and controlled power sequencing is fine for normal operation, of course.
In some power failure situations, room lights will flash on and off a few times as the power is going down and or when its coming back up. That transient pulsating power condition may not sync up well with a power sequencing circuit.
That's why I went for more of a simple brute force type solution.
In some power failure situations, room lights will flash on and off a few times as the power is going down and or when its coming back up. That transient pulsating power condition may not sync up well with a power sequencing circuit.
That's why I went for more of a simple brute force type solution.
Modern zener diodes have much lower reverse current. E.g. PLVA665A (6.5V zener) has max reverse current of 50nA at 80% Vz (i.e. 5.2V). Another option is to use a TVS diode.
Besides, all talk about possible performance degradation is speculation. The zener is applied to a non-critical DC power supply.
In my 4499 design I use separate digital and analog power supplies.
The analog have TPS7A39 regulator, its output is used for the OPs and for the LT3042, which provides 5V for the 4499.
All digital supplies are always ON, while TPS7A39 is ON in Idle/Play modes and is OFF in Standby.
Alex.
The analog have TPS7A39 regulator, its output is used for the OPs and for the LT3042, which provides 5V for the 4499.
All digital supplies are always ON, while TPS7A39 is ON in Idle/Play modes and is OFF in Standby.
Alex.