y1 gamma-1 DAC - Page 22

Could it also be the case that the DAC is expecting a high impedance at the output, but the 13 ohm impedance of the IEMs - almost a dead short - is causing noise problems?

The WM8501 DAC chip actually has stout enough an internal output stage to drive loads down to 16 ohms. However the γ1 was not designed to be a headphone amp, so its output coupling caps are too small for such low impedance loads (they form a high-pass filter with the load, leading to early bass rolloff). Assuming the stock 23uF output caps, γ1 can drive 300 ohm headphones (i.e., Sennheiser HD6x0) with fc of 23Hz, but with 16 ohm load the fc would rise to 432Hz -- no bass! With a 10K ohm load (i.e., the input of a Mini³), the fc is 0.7Hz.

At any rate, despite the lack of bass, a low impedance load will not "cause" noise.

On USB power, it measures 4.94 VDC with a 4.96 V supply from USB.

With the wallwart, it measures 5.15 VDC with a 5.17 VDC supply.

Check!

bperboy, so it looks like your U6D is still not regulating. It's letting just about the full voltage through... With your DMM on DC V mode, measure the voltage at the top of U6D pin 2. Be very careful with the probes! Don't slip and short anything or you risk blowing something. It should be 0V.

Hmm, this is quite odd.. measured from pin 2 to ground, got 0vdc, then measured from the test point again, and got 4.6vdc.

EDIT: Should one of the USB 5v test points measure 0v when mated with the dac board?

The "EP' version is what the people at TI refer to as a "HiRel Enhanced Product".

Advantages Over Standard Commercial-Off-The-Shelf Products are:

Controlled baseline (one assembly/test site, one wafer fabrication site)
Extended temperature performance (specific temperature ranges are stated on the
datasheets and may be as great as -55°C to 125°C)
Qualification pedigree
Customer notification of major product changes.
Assurance from TI that the device will perform to datasheet electrical specifications in
environments that require extended temperatures.
Nickel-Palladium-Gold or Tin-Lead lead finishes are standard.


(It's an easy and inexpensive way for OEMs to avoid upscreening COTS (Commercial-Off-The-Shelf ) plastic encapsulated microcircuits (PEMs) for use in military and aerospace applications)

It's possible that you still had a marginal joint at pin 2, and exerting pressure with your DMM probe on it made it "connect". Nevertheless, 4.6V is lower than spec now (for a TPS793475)...

No. They should both be about 5V. VCC will be very slightly lower than VBUS (while running on USB power) due to the voltage drop across the TPS2115A power mux chip, but the difference should be in the millivolts.

Nope, i do not have a 5v regulated psu. I'll probably have to get one from jaycar.

Has any Australians have any experience with the MP-3144 regulated power adapters sold by Jaycar?

OK, I cant resist...

ARGH. I was not careful, and accidently shorted pins 2 and 3 of U6D (I think). Both test points measure full voltage. I assume I must now replace both regulators. Anything else that is damaged?

Odd. The voltage regulators' pin 2 is ground and pin 3 is the ENABLE pin (which is connected to pin 1, the INPUT pin). If you had shorted pins 2 and 3, it shouldn't blow the regulators, but would cause the TPS2115A to go into current-limit protection, which should prevent any damage.

I think you should set the boards aside and start looking at the schematic diagram to try to understand the circuit a bit, at least around the power and voltage regulator sections a bit. Then, start ohming things out carefully while referencing the schematic to try to make sense of what's what. It really is fairly simple and shouldn't be difficult to debug.

Heads up! R20D value update

After some testing, MisterX and I have decided to change the value of the R20D resistor. It was originally 75 ohms, the new value is 47K ohms. If you have already built a γ1 with the old value, it is an optional (but recommended) update.

You should hopefully have some spare 47K resistors because the Panasonic, Xicon and Multicomp resistors from Digikey, Mouser and Farnell, respectively, are sold in multiples of 10.

Background:

This resistor terminates the CS8416's RXP0 input, which is used for the S/PDIF datastream from the PCM2707's DOUT pin. Originally it was thought that 75 ohms was the appropriate termination for this line, but the PCM2707's DOUT pin is not "standard" S/PDIF. It's an unbuffered 3.3Vpp TTL output, not intended to drive a long transmission line or a low impedance load such as a 75 ohm resistor. At 75 ohms, the "3.3Vpp" output got loaded down to around 1Vpp. This is still more than enough voltage to drive the CS8416's RXP0 input (which has an VIH threshold of 150mV typ.), but nevertheless it makes the PCM2707's DOUT work harder than it should (1Vpp into 75 ohms is about 13mA of output current).

Since the distance between the PCM2707 and the CS8416 is short and doesn't qualify as a transmission line, there is no need for line termination. However, a high value resistor helps to keep the S/PDIF pulse waveform clean. We settled on 47K as the new resistor value because does not load the PCM2707 down (the output voltage is now ~3.3Vpp and output current is now only 70uA), gives good S/PDIF pulse waveforms on the scope, and is a value that you should already have.

Note that this change should reduce the amount of fast current pulses drawn by the PCM2707 and result in a cleaner 3.3V supply rail on the USB board. I don't expect it to actually improve the "sound" or the customary metrics such as noise floor or distortion, but it is a good, simple update, so I encourage you all to make this change.

The parts list on the γ1 website has been updated to reflect this change.

A change would do you good ( with thanks to MisterX )

Okay, my regulator hasn't blown. Probably just needed to let the current-limit protection reset or something. I measure 3.3v again at that regulator. As for hte two 5v test points, I noticed that both measure ~5 vdc, one a bit lower than the other, when either usb only is plugged in or both usb and the wallwart is plugged in. The test point for J2U- pin 1 doesn't measure any voltage when the wallwart only is plugged in and not usb. Looking at teh schematic, this seems to make sense to me.

I have reflowed pins 1-3, measure ~0.2 ohms between pin 2 and ground, infinite resistance between Vin and Vout. Roughly a short through the inductor on the Vout side of things, and good solder joints on all the caps immediately surrounding that particular section of the circuit.

It seems to me that the only solution remaining is that the regulator is dead; I find that a bit hard to believe, as this is the second one I've put in. It's a possibility, and I do have a few more regulators, but I'd rather not replace it again.

I don't have any spare 47k resistors...... do you think am I better off leaving the 75R in and keeping DOUT loaded, or pulling the resistor and risking a lower quality SPDIF signal?