Sunday, July 31, 2011

Home Made PCB Exposure Unit

The following is the very first DIY stuff of mine, every 'JOKE' started from here.


Info captured from Internet

High-Speed Digital Design Online Newsletter: Vol. 2 Issue 22
Sang Cheol Lee writes:

I am engaged in developing a kind of set-top box. It must receive differential ECL (100K compatible) and then process it. However, I am not familiar with ECL level design. I [would like] to use a single power source (5V for VCC, 0V for GND) , so I prefer to use PECL level devices with TTL devices.

I would like to directly connect the differential ECL signal to the differential PECL device at connector point (first point) of receiver. Is this O.K., or should I use a specific level shifter circuitry or level transformer to connect them? Finally, would I need some special skill to design the connection?

Dr. Johnson replies:

Thanks for your interest in High-Speed Digital Design.

The ECL logic family was originally intended to be used with power supply voltages of 0 V and -5.2 V. The normal logic levels with ECL are:

V(OH) = -0.9 V

V(OL) = -1.7 V

The term PECL means we are using ECL logic with different power supply voltages.

The old 0-V pin connects to Vcc=+5V

The old -5.2-V pin connects to Gnd=0V

The chip is now being powered by 5.0 V +/- 10, instead of 5.2 V. Most ECL chips can tolerate this difference. The PECL logic voltage levels are:

V(OH) = Vcc - .9 V = 4.1 V nom.

V(OL) = Vcc - 1.7 V = 3.3 V nom.

Note that the PECL logic levels are now dependent on the Vcc level. As Vcc changes, the output levels change with Vcc. The common-mode input range of a PECL differential receiver will not tolerate true ECL levels (-0.9V and -1.7V).

When connecting true ECL to PECL, you will need a voltage translation.

If the data has equal numbers of ones and zeroes (for example, with a Manchester-ceded data sequence) then the level translation may be accomplished by sending the signal through a pair of DC-blocking capacitors (0.1 uF capacitors), and then re-biasing the receiver to its mid-range level. Other than this simple case, there is no good, simple way to accomplish the re-biasing.

Best regards,
Dr. Howard Johnson

Gigabit-Ethernet Optical Transceiver Research

HP/Agilent HFBR-53D5

Recommended Circuit on HFBR-53D5 Datasheet

HFBR-53D5 Circuit in Production


R3 68R ---> R201 49R9
R2 68R ---> R200 49R9
R4 ---->
R1 ---->
R11 270R ---> R199 274R
R10 270R ---> R198 274R

Power Requirements

VCC = 4.75 to 5.25 V
Icct = 85mA (typical), 120mA (max)

Optical Characteristics

Transmitter Output Optical Power: -9.5 ~ -4 dBm
Receiver Input Optical Power: -17 dBm ~ 0 dBm

This means we can connect the TX and RX directly without attenuator.

Signal Detection

On datasheet page 10:

If Signal Detect output is not used, leave it open-circtuied.

Signal Input/Output

The HP/Agilent HFBR-53D5 is a PECL based solution.

PECL is differential signaling provides noise cancellation, in the the graph taken from Wikipedia, we can realize how it works:

Vlow = 3.4V
Vhigh = 4.2V
Vcc = 5V


Most of the digital audio source signal is based on TTL, if we are going use the HP/Agilent HFBR-53D5 for the inter-chassis transmission, obviously the TTL needs to be converted to PECL on the TX side and then be converted back to TTL again on the RX side.

Another possible solution will be TTL based optical transceiver which supports 20Mbps or above, however I couldn't find such kind of transceiver.

Possible Solution for TTL/PECL Conversion

Here are some possible solutions :


[ To Be Continued ]

Alternative Transmission Consideration for Digital Audio

Just to post some photos of the optical transceiver for high speed network transmission here.

Pulse Transformer on 10/100 BaseTX

Optical Transceiver on OC3/ATM

Optical Transceiver on Gigabit-Ethernet

Optical Transceiver on OC48

XENPAK Optical Transceiver on 10GE

The Question on My Mind

What if we use those solution to replace the TOSLink?
What will happen? Interesting ................


Please Read the Specification of Your Scope

Many DIYer like myself, focused on the bandwidth of the oscilloscope itself.
However, when measuring the signal, the bandwidth of the probe will be another important variable needs to be taken into consideration.

The Probe Spec Example

If we take a look at the specification of the probe in the example above, we might notice that the probe only supports up to 6MHz at 1X position.

The Difference between 1X and 10X

Here's SPDIF signale measured with Probe at 1X

The following is the SPDIF measured with probe at 10X

As we can see, if we measure the signal with probe at 10X, the result gives us more clue for the noise.

Well, the old saying is right, RTFM.........

Friday, July 29, 2011

PO74G38072 Digital Buffer


Toshiba TC74VHCU04 is a well-known digital audio buffer chip since it was used in the famous DSIX circuit.
The PO74G38072 designed by Potato Semiconductor is a buffer chip for digital signal, supporting up to 1GHz. PO74G38072 was recommended by some DIYers as a alternative solution for digital buffer than the TC74VHCU04 due to the wide-bandwidth and its perfect specifications. I did a simple test, replacing the TC with PO chip, and the following is my discovery.

Listening difference

It was a surprise to me......... Surprise? So, better or worse?

Unfortunately, after swapping TC74VHCU04 with PO74G38072, my system sounds worse.
Actually, the two chips sounds nearly the same, however TC74VHCU04 provides more silent background than PO74G38072.
As a result, the TC74VHCU04 gives more dynamics in general.


I then tried to find out the reason technically.

Here's the comparison of the TC74VHCU04 and Po74G38072 on the oscilloscope:



Let's take a closer look:



As you can see, TC74VHCU04 has less ringing and overshoot than PO74G38072. In other words, less noise.
I think basically that's why TC74VHCU04 sounds better than PO74G38072 on my system.


I do believe that PO74G38072 should have better performance than TC74VHCU04, since it supports more bandwidth than TC. However, wider bandwidth sometimes means that the noise will passthrough the buffer as well. And I think that is basically why PO sounds worse than TC on my system. (Please note what I'm saying - my system)

But if I can decrease the noise on the power supply, I think there is more chance that the PO might sound better than TC for one reason - less jitter. PO is a much faster chip than TC, if we can well control the noise level sourced from the power supply, I do believe that the less jitter characteristics of the PO chip will be easily identified in this case.

I think the power supply will be the next task for me.