Cyton's default sample rate is 250 Hz. This is a limitation imposed by the RFduino Bluetooth radio link. Using a Wifi Shield with Cyton will allow faster rates. But the Shield is currently on hold for Cyton, while some power supply mods are tested. It will be back in the Shop store later this year. However even with the Shield, there are limitations in terms of ESP8266 and Wifi data rates.
To be realistic, most research EEG tops out at 500 or 1000 Hz. There is generally no need to go faster, as scalp EEG is greatly attenuated above 40 Hz gamma. In other words, increased sample rate above 500 or 1000 Hz, gets you NOTHING as far as increased resolution or information or signal quality.
I've actually gone over this in past threads with you here on the forum.
I am researching a special brain EEG, it is the sixthing feeling phenomenon. It like the psychological induction between the twin brother. When the phenomenon happen ,it must have a signal out from one people, and into other people's brain, what is signal, I think that it is cortex signal, cortex EEG. The cortex EEG can be 500HZ or higher. So I need the higher sample rate. The nervous excitation time is 0.5ms-2ms, so the sample rate need 4000HZ to 1000HZ. It is why I need the higher sample rate reason. My english is not good, I don't know whether you can understand my means. I have seen some very good detail on the 1-500HZ by cyton with WIFIshield, but it is not enough, so I hope you can offer 4000HZ sample rate or higher. Thank you so much.
You cannot record cortical EEG from the scalp. Requires removing scalp and bone to reach direct cortex surface. Can only be done by a neurosurgeon in a hospital. Scalp EEG stops not much higher than 40 Hz. Due to conduction losses of the bone and other tissues. Also there is considerable contamination of 40+ Hz EEG with EMG (muscle artifact) from scalp muscles. If you have ever looked at 40 Hz EEG amplitudes, they are very very low, only a few microvolts and lots of noise.
On most Wireless networks, the WiFi Shield can stream well at 1000Hz with the Cyton and Daisy boards and at 1600Hz with the Ganglion board. With a high speed network switch, sample rates over 2000Hz are possible with a theoretical limit of 16000Hz for the Cyton, 8000Hz for the Cyton with Daisy shield, and 12800Hz for the Ganglion.
So unfortunately the 'theoretical limit' mentioned is just that, not possible with actual wifi equipment. The practical achieved rates of around 1K to 2K are more in line with reality. The Wifi Shield will be back in the Shop later this year. It's undergoing some power supply circuit adjustments to better work with Cyton. It has always worked fine with Ganglion.
It may be possible to sample at higher rates to the SD card, but here again, there are limitations involving SD card buffering and write speeds.
More past Forum threads related to SD card sample rates here,
I have good plan maybe it can solve the problem of raising the sample rate to 2000HZ or higher. I fond the EEG signal of the two adjacent electrode is often the same. So if you can control the different channel start time, we can make one channel start at 0ms and other channel start time at the 0.05ms ( the sample rate is 1000HZ). Because the two channel (adjacent) have different start time ,the two signal will also be different. One sample point is 1ms,2ms,3ms......, other one is 0.5ms, 1.5ms, 2.5ms, if we add the two signal, the sample point will be 0.5ms, 1ms, 1.5ms, 2ms, 2.5ms, it will be 2000HZ sample rate point, so we will get the 2000HZ sample rate by add the two signal. Use this way, we also get 3000Hz sample rate, the sample point will be 0.3ms,0.6ms,0.9ms,1.2ms...., the precondition is you can control the different channel start time, I don't know you whether you can do it. can you try it?
What you mention is not the way analog to digital converters ADCs work, they sample at a constant frequency. The limitation with Cyton is primarily the 'bandwidth' (bits per second throughput) of the Bluetooth or Wifi (ESP8266) channel and constraints. You are limited in the amount of data you can push through the 'pipe', by the speed of the stream through the pipe (radios and protocols.)
All channels must sample at the same time, exactly, so that correlations such as coherence, phase, 3D localizations, etc., can be made by signal processing the data stream shown at the same time instant across all channels.
Nice workaround Profeell, however, not feasible.
Thinking about using the wired output and storing directly to USB, as I need to record at least 100 samples in 10 ms (SR: 10KHz).
I am not sure about the feasibility or availability of the much-awaited WiFI shield.
However, any suggestion would be great!
Thank your reply, I have other one plan, the single electrode must be analog signal, if we set 8 wires on one electrode and the 8 wires will connect 8 different channel, then we make them sample rate start time is different, for exmple: first chanel start time is 0.125ms, second start time is 0.25ms, the third start time is 0.375ms ......the eighth channel is 0.975ms, so will get 8000HZ sample rate point, next step is add the 8 channel data, we will get 8000HZ single channel data, I think that it can be availability. My English is no good, I wish you can understand my means. This plan need SDK support for set the different channel start time.
Thank !!
add, I don't change everyone channel sample rate, they are still 1000HZ, it also doesn't raise the WIFI bandwidth, your 'openbci GUI' have been given the sample rate 1000HZ( cyton board with WiFi board), the cyton can get 8000HZ sample rate on theory, so it can make 8000 Sample point ,I only want to use 8 channel record the 8000 sample data point, and every channel record 1000 data point per second( =1000HZ), the time poing is different, and the 1000HZ sample rate time interval is not change.
Thank you.
...for example: first channel start time is 0.125ms, second start time is 0.25ms, the third start time is 0.375ms...
I mentioned previously that the ADS1299 ADC chip does NOT work like this. All channels are sampled simultaneously. This is a requirement of ALL ADC chips.
There is still no timeline on the Shield availability. However the engineering group is now organizing resources to prioritize the hardware change details. You should note that most engineering staff were busy with Galea in 2021. Which put the Shield at a lower priority.
re: 2000 Hz EEG samples, please see this previous thread and comment. EEG scalp brainwaves above 50 Hz or so, are greatly attenuated by the skull and EMG contamination. So 2000 Hz sample rate does NOT achieve more resolution.
Dear all,
Just a thought:
Can we use ADS126x, as it offers nV resolution with a 38kSPS data rate?
This sounds exciting, however, would love here your point of view.
Thanks, in advance!
Regards,
Rathin
Comments
Profeel, hi.
Cyton's default sample rate is 250 Hz. This is a limitation imposed by the RFduino Bluetooth radio link. Using a Wifi Shield with Cyton will allow faster rates. But the Shield is currently on hold for Cyton, while some power supply mods are tested. It will be back in the Shop store later this year. However even with the Shield, there are limitations in terms of ESP8266 and Wifi data rates.
To be realistic, most research EEG tops out at 500 or 1000 Hz. There is generally no need to go faster, as scalp EEG is greatly attenuated above 40 Hz gamma. In other words, increased sample rate above 500 or 1000 Hz, gets you NOTHING as far as increased resolution or information or signal quality.
I've actually gone over this in past threads with you here on the forum.
https://openbci.com/forum/index.php?p=/discussion/1642/cannot-get-500-or-1000-hz-sample-rate-from-wifi-shield-or-sd-card
Etc.
William
I am researching a special brain EEG, it is the sixthing feeling phenomenon. It like the psychological induction between the twin brother. When the phenomenon happen ,it must have a signal out from one people, and into other people's brain, what is signal, I think that it is cortex signal, cortex EEG. The cortex EEG can be 500HZ or higher. So I need the higher sample rate. The nervous excitation time is 0.5ms-2ms, so the sample rate need 4000HZ to 1000HZ. It is why I need the higher sample rate reason. My english is not good, I don't know whether you can understand my means. I have seen some very good detail on the 1-500HZ by cyton with WIFIshield, but it is not enough, so I hope you can offer 4000HZ sample rate or higher. Thank you so much.
You cannot record cortical EEG from the scalp. Requires removing scalp and bone to reach direct cortex surface. Can only be done by a neurosurgeon in a hospital. Scalp EEG stops not much higher than 40 Hz. Due to conduction losses of the bone and other tissues. Also there is considerable contamination of 40+ Hz EEG with EMG (muscle artifact) from scalp muscles. If you have ever looked at 40 Hz EEG amplitudes, they are very very low, only a few microvolts and lots of noise.
https://www.frontiersin.org/articles/10.3389/fnhum.2013.00138/full
"High-frequency brain activity and muscle artifacts in MEG/EEG: a review and recommendations"
I am working on Brainstem Response Acquisition where the entire game is of 10 ms.
I need at least 100 samples in 10 ms. Hence, at least 10 kHz is desirable.
WiFi Shield claimed to support till 16kHz. Any idea/alternatives to acquired brainwaves at such a higher data rate?
Thank you.
@Rathin63, hi.
The actual Shop page snapshot from 2018 is here,
https://web.archive.org/web/20180822092044/https://shop.openbci.com/collections/frontpage/products/wifi-shield?variant=44534009550
https://github.com/OpenBCI/OpenBCI_WIFI
It stated:
So unfortunately the 'theoretical limit' mentioned is just that, not possible with actual wifi equipment. The practical achieved rates of around 1K to 2K are more in line with reality. The Wifi Shield will be back in the Shop later this year. It's undergoing some power supply circuit adjustments to better work with Cyton. It has always worked fine with Ganglion.
It may be possible to sample at higher rates to the SD card, but here again, there are limitations involving SD card buffering and write speeds.
More past Forum threads related to SD card sample rates here,
https://www.google.com/search?q=sd+card+sample+rate+site:openbci.com
Regards, William
I have good plan maybe it can solve the problem of raising the sample rate to 2000HZ or higher. I fond the EEG signal of the two adjacent electrode is often the same. So if you can control the different channel start time, we can make one channel start at 0ms and other channel start time at the 0.05ms ( the sample rate is 1000HZ). Because the two channel (adjacent) have different start time ,the two signal will also be different. One sample point is 1ms,2ms,3ms......, other one is 0.5ms, 1.5ms, 2.5ms, if we add the two signal, the sample point will be 0.5ms, 1ms, 1.5ms, 2ms, 2.5ms, it will be 2000HZ sample rate point, so we will get the 2000HZ sample rate by add the two signal. Use this way, we also get 3000Hz sample rate, the sample point will be 0.3ms,0.6ms,0.9ms,1.2ms...., the precondition is you can control the different channel start time, I don't know you whether you can do it. can you try it?
Profeell, hi.
What you mention is not the way analog to digital converters ADCs work, they sample at a constant frequency. The limitation with Cyton is primarily the 'bandwidth' (bits per second throughput) of the Bluetooth or Wifi (ESP8266) channel and constraints. You are limited in the amount of data you can push through the 'pipe', by the speed of the stream through the pipe (radios and protocols.)
All channels must sample at the same time, exactly, so that correlations such as coherence, phase, 3D localizations, etc., can be made by signal processing the data stream shown at the same time instant across all channels.
Regards, William
Nice workaround Profeell, however, not feasible.
Thinking about using the wired output and storing directly to USB, as I need to record at least 100 samples in 10 ms (SR: 10KHz).
I am not sure about the feasibility or availability of the much-awaited WiFI shield.
However, any suggestion would be great!
Cheers,
Rathin
Thank your reply, I have other one plan, the single electrode must be analog signal, if we set 8 wires on one electrode and the 8 wires will connect 8 different channel, then we make them sample rate start time is different, for exmple: first chanel start time is 0.125ms, second start time is 0.25ms, the third start time is 0.375ms ......the eighth channel is 0.975ms, so will get 8000HZ sample rate point, next step is add the 8 channel data, we will get 8000HZ single channel data, I think that it can be availability. My English is no good, I wish you can understand my means. This plan need SDK support for set the different channel start time.
Thank !!
add, I don't change everyone channel sample rate, they are still 1000HZ, it also doesn't raise the WIFI bandwidth, your 'openbci GUI' have been given the sample rate 1000HZ( cyton board with WiFi board), the cyton can get 8000HZ sample rate on theory, so it can make 8000 Sample point ,I only want to use 8 channel record the 8000 sample data point, and every channel record 1000 data point per second( =1000HZ), the time poing is different, and the 1000HZ sample rate time interval is not change.
Thank you.
I mentioned previously that the ADS1299 ADC chip does NOT work like this. All channels are sampled simultaneously. This is a requirement of ALL ADC chips.
Can I ask a question? When will the WIFI shield board back shop? If it can be back the shop, the sample rate will get the 2000HZ?
There is still no timeline on the Shield availability. However the engineering group is now organizing resources to prioritize the hardware change details. You should note that most engineering staff were busy with Galea in 2021. Which put the Shield at a lower priority.
re: 2000 Hz EEG samples, please see this previous thread and comment. EEG scalp brainwaves above 50 Hz or so, are greatly attenuated by the skull and EMG contamination. So 2000 Hz sample rate does NOT achieve more resolution.
https://openbci.com/forum/index.php?p=/discussion/comment/10330/#Comment_10330
Regards, William
Dear all,
Just a thought:
Can we use ADS126x, as it offers nV resolution with a 38kSPS data rate?
This sounds exciting, however, would love here your point of view.
Thanks, in advance!
Regards,
Rathin