Cheap 10 MHz distribution amplifier - caveat emptor

[dnessett]

Recently, I purchased an inexpensive 10 MHz distribution amplifier on Ebay (fig. 1 - NB: the penny is there to give an indication of size. For those unfamiliar with US pennies, they are about 18 mm in diameter). I tested the unit and discovered it had a defect (channel 2 output a 430 mV sine wave, whereas all others output a 3.3V sine wave). During the process of investigating what was the problem in consultation with the seller, I opened it up and took some photos. Ultimately, the seller and I could not come to an agreement how much discount of the original price I should receive to fix the amplifier (which was advertised as brand new), so I shipped it back to him for a refund.

Figure 1

However, I thought readers of this forum might be interested in the characteristics of the unit and what it looks like inside the case. Fig. 2 shows the input to the amplifier (yellow trace) and the output signal on channel 1 (blue trace). Notice that the input is 1.5 V P-P and there is a phase shift between the input and output of 18 ns. Since the period of a 10 MHz signal is 100 ns, the phase shift is about 65 degrees. This shift is probably due to delays through signal preprocessing components, including an inverting buffer suitable for driving a 50 ohm load (see below).

Figure 2

Fig. 3 shows the same two signals with the measurement cursors bracketing the output signal, which is 3.62V P-P.

Figure 3

Fig. 4 shows the phase shift on channel 5, which is 17.6 ns. This represents 63 degrees of phase shift. Consequently, the phase shift varies by several degrees on each channel, even when those channels are roughly in phase.

Figure 4

Fig. 5 shows the phase shift on channel 3, which is 67.8 ns. This is 244 degrees of phase shift. It appears that channel 3 is 180 degrees out of phase with an additional 65 degrees due to delays through the inverting buffer.

Figure 5

During the negotiation between the seller and me, I obtained from him a partial schematic of the board. Since I do not have his permission to post that image, I have reproduced some of it using LTSpice. This is shown in figure 6 (ignore the component identifiers, since they do not correspond with the board identifiers, which are shown in the next figure).

Figure 6

Fig. 7 shows the bottom of the PC board. The input sine wave arrives on the left side, is filtered then converted from a sine wave to a square wave (according to glarrson post) and then fed to a 74LVC2G04 buffer. The output of the buffer is filtered to obtain a sine wave and the result presented at the BNC connector of the associated channel.

Figure 7

The schematic is fairly unremarkable, the only somewhat interesting point being the use of the 74LVC2G04, which sources a high current signal sufficient to drive a 50 ohm load. There is nothing in the schematic that suggests why 4 channels are over 180 degrees out of phase with the other 4 channels. I attempted to trace the signal on the photo of the PC board bottom from input to output channels in order to identify the reason for the phase shift, but was unsuccessful.

[cdev]

That is kind of interesting. In other words, its as if some time was lost, where did it go?

You I'm sure didnt see any differences in the inductors, etc.

So anyway, this is kind of interesting, I agree.

This may or may not be helpful, but as I kind of am curious now.

I just recently bought a used extron video distribution amplifier which I plan to use for 10 MHz.


 I'd be happy to see if the output of signals fed into my new video distribution amplifier get phase shifted differentially and if so how much.. using my 2200 series Tek scope and my cheap feeltech signal generator which despite its cheapness can generate two signals that are out of phase with one another by arbitrary amounts

The video distrubution amps seem like the way to go for this.

I spent under $20 for it. When I was looking for it I saw another similar unit for around the same price but the listing wasn't set up to allow me to make an offer on it so I passed on that one. Thankfully I did because a few minutes later this one popped up.

The case is almost completely empty so I am going to use it for my GPSDO and a controller board and its display. There will still be more room after that is done. I paid under $20 for it.

 The point I am trying to make is they seem often to be available for very little and might end up being quite good for use to distribute 10 MHz and 1PPS  It looks to me as if I should be able to get 9 10 Mhz outputs and 3 PPS outputs for my GPSDO, plus possibly have four additional BNCs that I may be able to use for something else.   

[dnessett]

That is kind of interesting. In other words, its as if some time was lost, where did it go?

The filtering will introduce phase shifts. The variation between the two outputs roughly in phase is only .4 ns, which is probably due to variations in the two 74LVC2G04 devices. What puzzles me is why the designer decided to provide two banks of output, each bank being ~180 degrees out of phase.

[cdev]

Its my understanding that when distributing 10 MHz for high precision timing its supposed to be isolated using a transformer/balun like Ethernet is.

[dnessett]

Its my understanding that when distributing 10 MHz for high precision timing its supposed to be isolated using a transformer/balun like Ethernet is.

Probably. But remember, this is a low cost distribution amplifier (less than $100). For that price, it is not really intended for high precision timing. It is for the hobbiest, not the professional.

[cdev]

(I read that in a Time Nuts list posting)


There really isn't any reason why a 10 MHz amp has to be expensive.

[David Hess]

Its my understanding that when distributing 10 MHz for high precision timing its supposed to be isolated using a transformer/balun like Ethernet is.

Isolation prevents ground loops but potentially adds more temperature dependent phase shift.

The filtering will introduce phase shifts. The variation between the two outputs roughly in phase is only .4 ns, which is probably due to variations in the two 74LVC2G04 devices. What puzzles me is why the designer decided to provide two banks of output, each bank being ~180 degrees out of phase.

The phase shifts are not necessarily a problem if fixed, but strong filtering will certainly add a large temperature dependent phase shift which may not be acceptable.

Having two banks of outputs which are 180 degrees out of phase is apparently a pretty common feature for timing distribution amplifiers.  I assume this is to allow for a differential clock signal or two phase clocking.

[cdev]

We should never see any devices that act like a transmission line or delay line unless they are intended to do something like that- at least that is my understanding.

However, I have to say, it always makes my ears perk up when I am tuning around the shortwave bands, for example, and hear echoes long enough to be discernible.

"Probably a man made and not a natural phenomenon". 

But sometimes it does indeed seem natural.

[dnessett]

Having two banks of outputs which are 180 degrees out of phase is apparently a pretty common feature for timing distribution amplifiers.  I assume this is to allow for a differential clock signal or two phase clocking.

Interesting. I have been looking at used HP 5087A distribution amplifiers. Unfortunately, there are 4 configurations, only one of which routes a 10 MHz signal to all outputs. So far, I haven't been able to find one of these. However, I looked at the 5087A operation manual and as far as I can tell no 180 degree phase shift exists between any outputs. What instrument, other than the cheap one we have been discussing, has dual banks each 180 degrees out of phase with each other?

[nfmax]

Correct me if I am wrong, but isn't it a bad idea to use a limiting amplifier such as a digital inverter? The output amplitude depends on the supply rail, which means that any noise, hum, ripple etc on the supply will appear as added AM sidebands around the 10MHz carrier.

[xrunner]

Recently, I purchased an inexpensive 10 MHz distribution amplifier on Ebay (fig. 1 - NB: the penny is there to give an indication of size. For those unfamiliar with US pennies, they are about 18 mm in diameter). I tested the unit and discovered it had a defect (channel 2 output a 430 mV sine wave, whereas all others output a 3.3V sine wave).

I've had the same one but with an older date (2015-01-06). I checked my outputs after reading this post but they look OK. I added an on/off switch myself, instead of pulling the power plug. I don't want it on except when I need it for an instrument (I do leave my GPSDO on all the time though).

[TheSteve]

I have an HP/Symmetricom 58502A for my 10 MHz distribution. All outputs have a common ground which seems to be normal with commercial reference distribution hardware.

[cdev]

Where do you get your 50 ohm terminators - are they affordable?

[xrunner]

Where do you get your 50 ohm terminators - are they affordable?

I got them several years ago off Ebay, so I do not remember the price, but they were affordable in my mind. I see some on Ebay now - $12.95 for 8 pcs, better deals from China suppliers.

[Tom45]

I got them several years ago off Ebay, so I do not remember the price, but they were affordable in my mind. I see some on Ebay now - $12.95 for 8 pcs, better deals from China suppliers.

Do you find it necessary to terminate the unused outputs on your distribution amplifier?

[David Hess]

Having two banks of outputs which are 180 degrees out of phase is apparently a pretty common feature for timing distribution amplifiers.  I assume this is to allow for a differential clock signal or two phase clocking.

Interesting. I have been looking at used HP 5087A distribution amplifiers. Unfortunately, there are 4 configurations, only one of which routes a 10 MHz signal to all outputs. So far, I haven't been able to find one of these. However, I looked at the 5087A operation manual and as far as I can tell no 180 degree phase shift exists between any outputs. What instrument, other than the cheap one we have been discussing, has dual banks each 180 degrees out of phase with each other?

I did not have any specific 10 MHz distribution amplifier in mind.  I just occasionally run across this configuration.

I do not know of any test instruments which can take advantage of it but a custom circuit certainly could.

[edpalmer42]

You can kind of group BNC 50 ohm terminators into 3 classes:

1. 10Base2 Ethernet Thinnet terminators:  Cheap, fine for 10 MHz, test if you want to use them at higher frequencies.

2. Chinese clones:  Cheap, probably okay for 10 MHz but test them, not sure I'd want to try them for high frequencies, YMMV.

3. Name brand units, e.g. Amphenol, Minicircuits, Pomona, Pasternack:  Often expensive, always check datasheets.  Even the big companies like Amphenol and Pomona make lower quality products.

I've got a random collection of BNC terminators.  At 100 MHz, they all have a Return Loss of > 30 dB (SWR of < 1.06).  At 1 GHz, I see 8 - 18 dB (SWR of 2.3 - 1.3).  For comparison, I've also got a random batch of unlabelled SMA terminators.  At 1 GHz, they all measure 28 - 30 dB.  I don't know if those numbers are valid or if my test equipment is the limiting factor.

I don't know if it's necessary, but I tend to terminate most of my unused 10 MHz outputs on equipment just so I know they're not interfering.  I don't do that with my distribution amp because since the outputs are in phase, they shouldn't be a problem ...... right? 

Ed

[David Hess]

I don't know if it's necessary, but I tend to terminate most of my unused 10 MHz outputs on equipment just so I know they're not interfering.  I don't do that with my distribution amp because since the outputs are in phase, they shouldn't be a problem ...... right? 

There is also no transmission line hanging off the back of each unused output.

A good distribution amplifier design should have so much isolation between the inputs that termination is not needed under any circumstances except as the receiver requires.

[edpalmer42]

I don't know if it's necessary, but I tend to terminate most of my unused 10 MHz outputs on equipment just so I know they're not interfering.  I don't do that with my distribution amp because since the outputs are in phase, they shouldn't be a problem ...... right? 

There is also no transmission line hanging off the back of each unused output.

A good distribution amplifier design should have so much isolation between the inputs that termination is not needed under any circumstances except as the receiver requires.

It depends what you're doing.

Since I'm a time-nut, I routinely make measurements in the 10e-12 range, sometimes lower.  At those levels, interference can come from anywhere - between outputs, ground loops, coupling between two cables that happen to be close to each other, etc.  I swear that sometimes it seems that negative thoughts affect the measurements!

Ed

[dnessett]

Correct me if I am wrong, but isn't it a bad idea to use a limiting amplifier such as a digital inverter? The output amplitude depends on the supply rail, which means that any noise, hum, ripple etc on the supply will appear as added AM sidebands around the 10MHz carrier.

The power supply + pin is pretty heavily filtered on the 74LVC2G04 chip (see the left hand side of the schematic). I don't know if it is "enough". I didn't have a chance to look at the spectrum of an output before I shipped the unit back to the seller. Perhaps someone who currently has a unit might do this and post the results.

[xrunner]

I didn't have a chance to look at the spectrum of an output before I shipped the unit back to the seller. Perhaps someone who currently has a unit might do this and post the results.

Here's mine -

[dnessett]

I didn't have a chance to look at the spectrum of an output before I shipped the unit back to the seller. Perhaps someone who currently has a unit might do this and post the results.

Here's mine -

Looks reasonable to me. The reason for the 3rd harmonic magnitude exceeding the 2nd is probably due to the conversion of the input sine wave to a square wave (according to glarrson) and then filtering back to a sine wave.

[jschatzman]

I bought one of these and it worked for about a year and then died. The signal-detect LED lights but there are no outputs on any of the output lines. I haven't tried to diagnose the problem. I think that I am going to stick with the SRS FS710, of which I have several, and which seem to work well.

[ArthurDent]

The problem with buying a video amp is there are so many variations, even from the same company, that you never know what the circuit looks like until you buy one and open it up. I bought one (RGBS) with lots of outputs but found that although the circuit board was set up for a separate opamp for each output on the channel, that only one opamp was used and the other outputs on that channel were just jumpered so one opamp was driving all 6 outputs on that channel. Also, the opamp used didn’t have the necessary bandwidth and would start clipping peaks of the 10 Mhz sinewave at 1V.

What I did was find some better higher bandwidth opamps to populate all 6 outputs per channel and now the amp has high level undistorted 10 Mhz output and each channel is isolated from each other and the input.

One problem a lot of people have is removing and replacing the 75 ohm resistors for the input/output to convert the video amps to 50 ohms. A simpler solution is to just solder a 150 ohm resistor across each 75 ohm resistor and the equivalent resistance is 50 ohms. Easier and far less chance of damaging the board removing the 75 ohm resistors.

[ArthurDent]

David Partridge used an interesting and effective scheme to synchronize outputs on his divider board (no longer available) that could output several different frequencies that would all start at the same time. I have one of the boards he produced and it works great. Here is the description of the circuit that might be of help.

http://www.perdrix.co.uk/FrequencyDivider/