# Was the Soviet N1 really capable of sending 9.6 GB/s of telemetry?

On the Wikipedia page for Soviet N1, it says of the control system:

The telemetry system relayed data back at an estimated rate of 9.6 gigabytes per second on 320,000 channels on 14 frequencies. Commands could be sent to an ascending N1 at the same rate.

There was a source for that claim though, a book available on Google books. https://books.google.com/books?id=nVeY7vMCtOkC&pg=PA226#v=onepage&q&f=false

This send pretty unbelievable to me considering general technology of the time, and the similar page on the Saturn V mentions about 200 channels of telemetry and 2 or 3 different transmitting frequencies.

Does this make sense? If so, how was this achieved in the late 60s/early 70s?

• key words from the source: "some analysts estimating" Feb 21, 2019 at 18:04
• Note that these concerns were already raised on the article's talk page in December 2017. I've tagged it as dubious for now and will remove the statement in a few days if there are no objections. Feb 21, 2019 at 19:30
• @isanae thanks for pointing that out. I don't really participate over there, just read; it never would have occurred to me to check that. Feb 21, 2019 at 22:57
• Without encoding multiple bits per clock cycle (e.g. multiple amplitudes or phases) you would need a bandwidth of about one Hertz per bit/sec transmitted, in this case about 77 GHz. And the carrier frequency is typically chosen much larger than the bandwidth. Spatial multiplexing is probably out of question for communication to a single ground control station. 9.6 kBits/sec sounds more realistic... Feb 22, 2019 at 15:57
• Unless they 'sent' the data by ejecting a bunch of mag tapes with the telemetry written on it, this sounds completely implausible for the technology of the time. It's perhaps possible that they could have recorded telemetry very quickly, but actually sending it back to a ground station that fast is a completely different matter. Feb 22, 2019 at 16:38

That claim is rather dubious. First, there is the claim of 320,000 channels of telemetry, while one paragraph earlier it lists 13,000 sensors on board. There will be setpoints in addition to sensor data, but 20x as many?

The earlier 5L mission had 10,000 telemetry channels.

I found these specifications for the S-530 computer: speed: 0.1 MIPS
RAM: 256 13-bit words
ROM: 8,192 20-bit words
components: hybrid ICs Tropa
design: NII AP

I don't see how a computer with these specs could possibly generate Gbits/s of data.

Speculation: the 9.6 Gbit/s is an estimate by US intelligence analysts who listened in on the launch. Maybe they got something wrong and their recordings were garbled for a reason other than excessive data volume.

• [serious face] "Look at these specs I mad..., er, calculated. The enemy is far ahead of us. Give us more funding!". Feb 21, 2019 at 14:31
• "I don't see how a computer with these specs could possibly generate Gbits/s of data." If the sensors each have their own transmitter, it may not need much of a computer. I somehow suspect that's how they got the large amount of channels too.
– Mast
Feb 21, 2019 at 15:02
• @Contango The USSR also engaged in deception about such things. For example, at a parade in 1955, they flew ten brand new M-4 bombers in formation, which then looped around, joined up with eight more M-4s and flew past again. American observers concluded that the USSR had 28 M-4s, rather than just 18, so believed for a long time that they were being built very quickly and in huge numbers. Feb 21, 2019 at 18:51
• @JosephRogers I don't see the problem, you just need about 320,000 tape drives. Feb 21, 2019 at 19:07
• @DavidRicherby Messing with enemy intelligence was a lot of fun all around. You know where the name of the US's "SEAL Team Six" came from? They wanted the Soviets to think there were five more just like them. Feb 21, 2019 at 19:45

I suspect this confuses "bandwidth available" with "bandwidth that can be used concurrently at any one time". It certainly could not be processed at that rate by the ground based systems, let alone on-board.

• Can you explain the difference between “bandwidth available” and “bandwidth that can be used concurrently at any one time”? Feb 22, 2019 at 7:48
• @Holger Well there are lots of reasons why that might be the case. My guess would be something like the channel have some sort of TDM (time sharing) on one of the frequencies. So 9.6G is 30K * 320,000 channels, but only fractions adding up to 14 of these channels can be used at any one time. That's about 0.5MBPS. For perspective this is more, but the same order of magnitude as the Apollo missions. Feb 22, 2019 at 9:55
• That would still imply processing the data at the specified rate while the particular unit has an assigned time slice. Which contradicts your statement, “It certainly could not be processed at that rate”. When the communication happens at a higher rate than the end points can process the data, there must be something buffering and sending/receiving at the specified rate, which brings the entire discussion back to square one. What kind of hardware of the sixties ought to be able to do that. Feb 22, 2019 at 10:05
• @Holger We are getting technical now here, but not necessarily. The decision to process doesn't necessarily want to depend on the data. For example, everyone gets a 1ms slice every 22 seconds would require almost no processing. Feb 22, 2019 at 10:21
• But getting “a 1ms slice every 22 seconds” doesn’t make a communication having a 9.6GByte/s rate. Only actually sending with 9.6GByte/s speed during the 1ms does that. Which implies buffering 9.6MByte and sending them within the 1ms. Which brings us back to the question which sixties hardware could not do that and was available and affordable to have 320,000 of them. Feb 22, 2019 at 10:45

I don't have any original sources, but I'd say: 9.6 GB/s are possible, albeit not an accurate description. It's very likely that the largest amount of data was not transported digitally but as raw analog signals. This has the advantage that you can take the raw output of the sensors and send them to ground where they can be recorded and/or digitized later on.

The reasons are simple: having all 320k channels converted to digital data on-board the rocket would be a huge amount of electronics involved, especially if you aim for a high resolution in both amplitude and time. If you transmit analog signals, you can store them on any tape and re-analyze them in any detail you like.

As a comparison: Our good old analog TV used around 200 MBit/s of analog data for a single TV channel - all possible channels combined on cable TV were well above 10 GBit/s. These were never digitized (until the advent of digital flat screens), so there were no processors involved in transporting the data.

• Good old analog TV needed not more than 80 MBit/s.
– Uwe
Feb 22, 2019 at 9:40
• Video, science and engineering feeds are likely not in the same system as the telemetry. I couldn't find sources for the N1, but it was the case for the Apollo, and i can't see a good reason to have such a critical system carry video feed... Feb 22, 2019 at 10:01
• @ANone I used TV as a comparison, I don't claim that they used video feeds. Feb 22, 2019 at 10:03
• Of course, you can make up any bitrate for an analog transmission, as air is capable of carrying it, but it’s still dubious, when you know that the actual hardware available at that time would not be able to digitize that signal with that precision, or well, not even the analog technology had the necessary precision, so if they stored the analog signal unprocessed, to process it today with modern technology, it would not yield 9.6G/s. Feb 22, 2019 at 10:13
• @asdfex I see. My point is its not fair to assume that means the telemetry system is capable of "similar speeds". Analog encoded of analog signals are really just SNR limited in terms of theoretical bandwidth. It's very unfair to expect to compare that to telemetry data. If the question was could 9.6G of signal be coming off the N1, this might be the case (I still have doubts) but 9.6G of telemetry seems highly implausible unless your definition of telemetry gets silly. Feb 22, 2019 at 10:17

While the claim is dubious, it is not entirely implausible even though most likely somewhat exaggerated. I'm also pretty sure it should be bits instead of bytes.

For example, VLBI (Very Long Baseline Interferometry) projects at that era, late 60s early 70s, handled bit loads in order of $$10^8$$ bits i.e. 100s of megabits. This data was collected in relatively short period (naturally, processing was - and still is! – a wholly different issue).

Now, if you divide that 10 GB/s with 320,000 you get around 30 kB/s/channel, completely plausible data rate. Sure, 320,000 recorders sounds a lot, but you also have to consider that this was a project that took considerable share of Soviet economy, some estimates put the share of defense & space programs up to 10% of GNP during that era, or ~400 billion (2019) dollars. If we estimate that the lunar rocket took 5-10% of that, you'd have 20-40 billions per year to spend on hardware.

• The claim is that it actually sent data that fast, though, which is almost certainly false. Collecting data that fast to local media and sending it that fast to a ground station are two very different things. Feb 22, 2019 at 16:35
• VLBI data was written to magnetic tape locally, then collected where it would be analyzed. It was certainly not transmitted by radio in real time. Feb 22, 2019 at 17:14